Hugging Face's logo

henryen
/
OriGen_Fix

PEFT
Safetensors
Model card Files Community
henryen commited on
Commit
53724ef
·
1 Parent(s): abb6ded

initial model upload

Browse files
Files changed (10) hide show
  1. adapter_config.json +34 -0
  2. adapter_model.safetensors +3 -0
  3. optimizer.pt +3 -0
  4. rng_state.pth +3 -0
  5. scheduler.pt +3 -0
  6. special_tokens_map.json +23 -0
  7. tokenizer.json +0 -0
  8. tokenizer_config.json +146 -0
  9. trainer_state.json +3813 -0
  10. training_args.bin +3 -0
adapter_config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "deepseek-ai/deepseek-coder-7b-instruct-v1.5",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 32,
14
+ "lora_dropout": 0.05,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 32,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": [
23
+ "k_proj",
24
+ "down_proj",
25
+ "o_proj",
26
+ "up_proj",
27
+ "v_proj",
28
+ "q_proj",
29
+ "gate_proj"
30
+ ],
31
+ "task_type": "CAUSAL_LM",
32
+ "use_dora": false,
33
+ "use_rslora": false
34
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cdffeaa4d10dcb211d4f83e1288091a14874dd066b63faec8572c15c2b854b4a
3
+ size 299883760
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b28e0a124af9159222f4e359fb0b7d0fbfb004f45efa635fa71f09e55a61be8
3
+ size 600008958
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:55821a39e1f875d4d7b22d0b67ed95234df8dcc42a20dab472a5790c4594e9b2
3
+ size 14244
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba218197106d2e55d81eb53e28ec4fc5a8d6b0d8e1f9c6ca8b7a21271466428a
3
+ size 1064
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<｜begin▁of▁sentence｜>",
4
+ "lstrip": false,
5
+ "normalized": true,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<|EOT|>",
11
+ "lstrip": false,
12
+ "normalized": true,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<｜end▁of▁sentence｜>",
18
+ "lstrip": false,
19
+ "normalized": true,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,146 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "100000": {
6
+ "content": "<｜begin▁of▁sentence｜>",
7
+ "lstrip": false,
8
+ "normalized": true,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "100001": {
14
+ "content": "<｜end▁of▁sentence｜>",
15
+ "lstrip": false,
16
+ "normalized": true,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "100002": {
22
+ "content": "ø",
23
+ "lstrip": false,
24
+ "normalized": true,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": false
28
+ },
29
+ "100003": {
30
+ "content": "ö",
31
+ "lstrip": false,
32
+ "normalized": true,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": false
36
+ },
37
+ "100004": {
38
+ "content": "ú",
39
+ "lstrip": false,
40
+ "normalized": true,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": false
44
+ },
45
+ "100005": {
46
+ "content": "ÿ",
47
+ "lstrip": false,
48
+ "normalized": true,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": false
52
+ },
53
+ "100006": {
54
+ "content": "õ",
55
+ "lstrip": false,
56
+ "normalized": true,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": false
60
+ },
61
+ "100007": {
62
+ "content": "÷",
63
+ "lstrip": false,
64
+ "normalized": true,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": false
68
+ },
69
+ "100008": {
70
+ "content": "û",
71
+ "lstrip": false,
72
+ "normalized": true,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": false
76
+ },
77
+ "100009": {
78
+ "content": "ý",
79
+ "lstrip": false,
80
+ "normalized": true,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": false
84
+ },
85
+ "100010": {
86
+ "content": "À",
87
+ "lstrip": false,
88
+ "normalized": true,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": false
92
+ },
93
+ "100011": {
94
+ "content": "ù",
95
+ "lstrip": false,
96
+ "normalized": true,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": false
100
+ },
101
+ "100012": {
102
+ "content": "Á",
103
+ "lstrip": false,
104
+ "normalized": true,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": false
108
+ },
109
+ "100013": {
110
+ "content": "þ",
111
+ "lstrip": false,
112
+ "normalized": true,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": false
116
+ },
117
+ "100014": {
118
+ "content": "ü",
119
+ "lstrip": false,
120
+ "normalized": true,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "100015": {
126
+ "content": "<|EOT|>",
127
+ "lstrip": false,
128
+ "normalized": true,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": true
132
+ }
133
+ },
134
+ "bos_token": "<｜begin▁of▁sentence｜>",
135
+ "chat_template": "{% if not add_generation_prompt is defined %}\n{% set add_generation_prompt = false %}\n{% endif %}\n{%- set ns = namespace(found=false) -%}\n{%- for message in messages -%}\n {%- if message['role'] == 'system' -%}\n {%- set ns.found = true -%}\n {%- endif -%}\n{%- endfor -%}\n{{bos_token}}{%- if not ns.found -%}\n{{'You are an AI programming assistant, utilizing the Deepseek Coder model, developed by Deepseek Company, and you only answer questions related to computer science. For politically sensitive questions, security and privacy issues, and other non-computer science questions, you will refuse to answer\\n'}}\n{%- endif %}\n{%- for message in messages %}\n {%- if message['role'] == 'system' %}\n{{ message['content'] }}\n {%- else %}\n {%- if message['role'] == 'user' %}\n{{'### Instruction:\\n' + message['content'] + '\\n'}}\n {%- else %}\n{{'### Response:\\n' + message['content'] + '\\n<|EOT|>\\n'}}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{% if add_generation_prompt %}\n{{'### Response:'}}\n{% endif %}",
136
+ "clean_up_tokenization_spaces": false,
137
+ "eos_token": "<|EOT|>",
138
+ "legacy": true,
139
+ "model_max_length": 4096,
140
+ "pad_token": "<｜end▁of▁sentence｜>",
141
+ "padding_side": "right",
142
+ "sp_model_kwargs": {},
143
+ "tokenizer_class": "LlamaTokenizer",
144
+ "unk_token": null,
145
+ "use_default_system_prompt": false
146
+ }
trainer_state.json ADDED
@@ -0,0 +1,3813 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.07845821231603622,
3
+ "best_model_checkpoint": "./checkpoints/ckpt_8_12_r32_debug_fp16_lr5e-05_batch8/checkpoint-3000",
4
+ "epoch": 4.980286366466072,
5
+ "eval_steps": 125,
6
+ "global_step": 3000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.01,
13
+ "learning_rate": 2.747252747252747e-06,
14
+ "loss": 0.6187,
15
+ "step": 5
16
+ },
17
+ {
18
+ "epoch": 0.02,
19
+ "learning_rate": 5.494505494505494e-06,
20
+ "loss": 0.6047,
21
+ "step": 10
22
+ },
23
+ {
24
+ "epoch": 0.02,
25
+ "learning_rate": 8.241758241758243e-06,
26
+ "loss": 0.5203,
27
+ "step": 15
28
+ },
29
+ {
30
+ "epoch": 0.03,
31
+ "learning_rate": 1.0989010989010989e-05,
32
+ "loss": 0.4791,
33
+ "step": 20
34
+ },
35
+ {
36
+ "epoch": 0.04,
37
+ "learning_rate": 1.3736263736263738e-05,
38
+ "loss": 0.4216,
39
+ "step": 25
40
+ },
41
+ {
42
+ "epoch": 0.05,
43
+ "learning_rate": 1.6483516483516486e-05,
44
+ "loss": 0.3502,
45
+ "step": 30
46
+ },
47
+ {
48
+ "epoch": 0.06,
49
+ "learning_rate": 1.923076923076923e-05,
50
+ "loss": 0.29,
51
+ "step": 35
52
+ },
53
+ {
54
+ "epoch": 0.07,
55
+ "learning_rate": 2.1978021978021977e-05,
56
+ "loss": 0.2761,
57
+ "step": 40
58
+ },
59
+ {
60
+ "epoch": 0.07,
61
+ "learning_rate": 2.4725274725274727e-05,
62
+ "loss": 0.2568,
63
+ "step": 45
64
+ },
65
+ {
66
+ "epoch": 0.08,
67
+ "learning_rate": 2.7472527472527476e-05,
68
+ "loss": 0.2493,
69
+ "step": 50
70
+ },
71
+ {
72
+ "epoch": 0.09,
73
+ "learning_rate": 3.021978021978022e-05,
74
+ "loss": 0.2458,
75
+ "step": 55
76
+ },
77
+ {
78
+ "epoch": 0.1,
79
+ "learning_rate": 3.296703296703297e-05,
80
+ "loss": 0.2542,
81
+ "step": 60
82
+ },
83
+ {
84
+ "epoch": 0.11,
85
+ "learning_rate": 3.571428571428572e-05,
86
+ "loss": 0.2133,
87
+ "step": 65
88
+ },
89
+ {
90
+ "epoch": 0.12,
91
+ "learning_rate": 3.846153846153846e-05,
92
+ "loss": 0.2177,
93
+ "step": 70
94
+ },
95
+ {
96
+ "epoch": 0.12,
97
+ "learning_rate": 4.120879120879121e-05,
98
+ "loss": 0.2397,
99
+ "step": 75
100
+ },
101
+ {
102
+ "epoch": 0.13,
103
+ "learning_rate": 4.3956043956043955e-05,
104
+ "loss": 0.2287,
105
+ "step": 80
106
+ },
107
+ {
108
+ "epoch": 0.14,
109
+ "learning_rate": 4.670329670329671e-05,
110
+ "loss": 0.2279,
111
+ "step": 85
112
+ },
113
+ {
114
+ "epoch": 0.15,
115
+ "learning_rate": 4.945054945054945e-05,
116
+ "loss": 0.2375,
117
+ "step": 90
118
+ },
119
+ {
120
+ "epoch": 0.16,
121
+ "learning_rate": 4.999976833475735e-05,
122
+ "loss": 0.2392,
123
+ "step": 95
124
+ },
125
+ {
126
+ "epoch": 0.17,
127
+ "learning_rate": 4.999882720206755e-05,
128
+ "loss": 0.2276,
129
+ "step": 100
130
+ },
131
+ {
132
+ "epoch": 0.17,
133
+ "learning_rate": 4.999716215008542e-05,
134
+ "loss": 0.2269,
135
+ "step": 105
136
+ },
137
+ {
138
+ "epoch": 0.18,
139
+ "learning_rate": 4.999477322702776e-05,
140
+ "loss": 0.2174,
141
+ "step": 110
142
+ },
143
+ {
144
+ "epoch": 0.19,
145
+ "learning_rate": 4.99916605020733e-05,
146
+ "loss": 0.2315,
147
+ "step": 115
148
+ },
149
+ {
150
+ "epoch": 0.2,
151
+ "learning_rate": 4.9987824065360743e-05,
152
+ "loss": 0.2137,
153
+ "step": 120
154
+ },
155
+ {
156
+ "epoch": 0.21,
157
+ "learning_rate": 4.998326402798612e-05,
158
+ "loss": 0.2415,
159
+ "step": 125
160
+ },
161
+ {
162
+ "epoch": 0.21,
163
+ "eval_loss": 0.19619159400463104,
164
+ "eval_runtime": 6.7054,
165
+ "eval_samples_per_second": 14.913,
166
+ "eval_steps_per_second": 14.913,
167
+ "step": 125
168
+ },
169
+ {
170
+ "epoch": 0.22,
171
+ "learning_rate": 4.997798052199959e-05,
172
+ "loss": 0.2074,
173
+ "step": 130
174
+ },
175
+ {
176
+ "epoch": 0.22,
177
+ "learning_rate": 4.99719737004016e-05,
178
+ "loss": 0.2289,
179
+ "step": 135
180
+ },
181
+ {
182
+ "epoch": 0.23,
183
+ "learning_rate": 4.996524373713848e-05,
184
+ "loss": 0.2396,
185
+ "step": 140
186
+ },
187
+ {
188
+ "epoch": 0.24,
189
+ "learning_rate": 4.995779082709739e-05,
190
+ "loss": 0.2294,
191
+ "step": 145
192
+ },
193
+ {
194
+ "epoch": 0.25,
195
+ "learning_rate": 4.994961518610064e-05,
196
+ "loss": 0.229,
197
+ "step": 150
198
+ },
199
+ {
200
+ "epoch": 0.26,
201
+ "learning_rate": 4.994071705089953e-05,
202
+ "loss": 0.2269,
203
+ "step": 155
204
+ },
205
+ {
206
+ "epoch": 0.27,
207
+ "learning_rate": 4.993109667916742e-05,
208
+ "loss": 0.2172,
209
+ "step": 160
210
+ },
211
+ {
212
+ "epoch": 0.27,
213
+ "learning_rate": 4.992075434949227e-05,
214
+ "loss": 0.2143,
215
+ "step": 165
216
+ },
217
+ {
218
+ "epoch": 0.28,
219
+ "learning_rate": 4.990969036136862e-05,
220
+ "loss": 0.211,
221
+ "step": 170
222
+ },
223
+ {
224
+ "epoch": 0.29,
225
+ "learning_rate": 4.989790503518888e-05,
226
+ "loss": 0.2261,
227
+ "step": 175
228
+ },
229
+ {
230
+ "epoch": 0.3,
231
+ "learning_rate": 4.9885398712234054e-05,
232
+ "loss": 0.2182,
233
+ "step": 180
234
+ },
235
+ {
236
+ "epoch": 0.31,
237
+ "learning_rate": 4.987217175466387e-05,
238
+ "loss": 0.2304,
239
+ "step": 185
240
+ },
241
+ {
242
+ "epoch": 0.32,
243
+ "learning_rate": 4.9858224545506296e-05,
244
+ "loss": 0.2293,
245
+ "step": 190
246
+ },
247
+ {
248
+ "epoch": 0.32,
249
+ "learning_rate": 4.9843557488646416e-05,
250
+ "loss": 0.2252,
251
+ "step": 195
252
+ },
253
+ {
254
+ "epoch": 0.33,
255
+ "learning_rate": 4.9828171008814786e-05,
256
+ "loss": 0.2294,
257
+ "step": 200
258
+ },
259
+ {
260
+ "epoch": 0.34,
261
+ "learning_rate": 4.9812065551575077e-05,
262
+ "loss": 0.2137,
263
+ "step": 205
264
+ },
265
+ {
266
+ "epoch": 0.35,
267
+ "learning_rate": 4.979524158331123e-05,
268
+ "loss": 0.1993,
269
+ "step": 210
270
+ },
271
+ {
272
+ "epoch": 0.36,
273
+ "learning_rate": 4.97776995912139e-05,
274
+ "loss": 0.2004,
275
+ "step": 215
276
+ },
277
+ {
278
+ "epoch": 0.37,
279
+ "learning_rate": 4.975944008326639e-05,
280
+ "loss": 0.2284,
281
+ "step": 220
282
+ },
283
+ {
284
+ "epoch": 0.37,
285
+ "learning_rate": 4.974046358822988e-05,
286
+ "loss": 0.2117,
287
+ "step": 225
288
+ },
289
+ {
290
+ "epoch": 0.38,
291
+ "learning_rate": 4.972077065562821e-05,
292
+ "loss": 0.2229,
293
+ "step": 230
294
+ },
295
+ {
296
+ "epoch": 0.39,
297
+ "learning_rate": 4.9700361855731883e-05,
298
+ "loss": 0.2124,
299
+ "step": 235
300
+ },
301
+ {
302
+ "epoch": 0.4,
303
+ "learning_rate": 4.967923777954158e-05,
304
+ "loss": 0.22,
305
+ "step": 240
306
+ },
307
+ {
308
+ "epoch": 0.41,
309
+ "learning_rate": 4.965739903877104e-05,
310
+ "loss": 0.2345,
311
+ "step": 245
312
+ },
313
+ {
314
+ "epoch": 0.42,
315
+ "learning_rate": 4.963484626582937e-05,
316
+ "loss": 0.2077,
317
+ "step": 250
318
+ },
319
+ {
320
+ "epoch": 0.42,
321
+ "eval_loss": 0.18103449046611786,
322
+ "eval_runtime": 6.668,
323
+ "eval_samples_per_second": 14.997,
324
+ "eval_steps_per_second": 14.997,
325
+ "step": 250
326
+ },
327
+ {
328
+ "epoch": 0.42,
329
+ "learning_rate": 4.96115801138027e-05,
330
+ "loss": 0.2069,
331
+ "step": 255
332
+ },
333
+ {
334
+ "epoch": 0.43,
335
+ "learning_rate": 4.95876012564353e-05,
336
+ "loss": 0.226,
337
+ "step": 260
338
+ },
339
+ {
340
+ "epoch": 0.44,
341
+ "learning_rate": 4.956291038811003e-05,
342
+ "loss": 0.2113,
343
+ "step": 265
344
+ },
345
+ {
346
+ "epoch": 0.45,
347
+ "learning_rate": 4.9537508223828266e-05,
348
+ "loss": 0.2033,
349
+ "step": 270
350
+ },
351
+ {
352
+ "epoch": 0.46,
353
+ "learning_rate": 4.9511395499189206e-05,
354
+ "loss": 0.2263,
355
+ "step": 275
356
+ },
357
+ {
358
+ "epoch": 0.46,
359
+ "learning_rate": 4.9484572970368514e-05,
360
+ "loss": 0.2187,
361
+ "step": 280
362
+ },
363
+ {
364
+ "epoch": 0.47,
365
+ "learning_rate": 4.945704141409648e-05,
366
+ "loss": 0.2176,
367
+ "step": 285
368
+ },
369
+ {
370
+ "epoch": 0.48,
371
+ "learning_rate": 4.942880162763549e-05,
372
+ "loss": 0.2297,
373
+ "step": 290
374
+ },
375
+ {
376
+ "epoch": 0.49,
377
+ "learning_rate": 4.939985442875695e-05,
378
+ "loss": 0.2101,
379
+ "step": 295
380
+ },
381
+ {
382
+ "epoch": 0.5,
383
+ "learning_rate": 4.937020065571763e-05,
384
+ "loss": 0.2089,
385
+ "step": 300
386
+ },
387
+ {
388
+ "epoch": 0.51,
389
+ "learning_rate": 4.9339841167235336e-05,
390
+ "loss": 0.1906,
391
+ "step": 305
392
+ },
393
+ {
394
+ "epoch": 0.51,
395
+ "learning_rate": 4.93087768424641e-05,
396
+ "loss": 0.222,
397
+ "step": 310
398
+ },
399
+ {
400
+ "epoch": 0.52,
401
+ "learning_rate": 4.927700858096866e-05,
402
+ "loss": 0.2094,
403
+ "step": 315
404
+ },
405
+ {
406
+ "epoch": 0.53,
407
+ "learning_rate": 4.924453730269849e-05,
408
+ "loss": 0.238,
409
+ "step": 320
410
+ },
411
+ {
412
+ "epoch": 0.54,
413
+ "learning_rate": 4.9211363947961083e-05,
414
+ "loss": 0.2104,
415
+ "step": 325
416
+ },
417
+ {
418
+ "epoch": 0.55,
419
+ "learning_rate": 4.917748947739478e-05,
420
+ "loss": 0.2025,
421
+ "step": 330
422
+ },
423
+ {
424
+ "epoch": 0.56,
425
+ "learning_rate": 4.9142914871940885e-05,
426
+ "loss": 0.2037,
427
+ "step": 335
428
+ },
429
+ {
430
+ "epoch": 0.56,
431
+ "learning_rate": 4.9107641132815325e-05,
432
+ "loss": 0.2179,
433
+ "step": 340
434
+ },
435
+ {
436
+ "epoch": 0.57,
437
+ "learning_rate": 4.907166928147965e-05,
438
+ "loss": 0.2265,
439
+ "step": 345
440
+ },
441
+ {
442
+ "epoch": 0.58,
443
+ "learning_rate": 4.9035000359611396e-05,
444
+ "loss": 0.2204,
445
+ "step": 350
446
+ },
447
+ {
448
+ "epoch": 0.59,
449
+ "learning_rate": 4.899763542907399e-05,
450
+ "loss": 0.2114,
451
+ "step": 355
452
+ },
453
+ {
454
+ "epoch": 0.6,
455
+ "learning_rate": 4.895957557188593e-05,
456
+ "loss": 0.2205,
457
+ "step": 360
458
+ },
459
+ {
460
+ "epoch": 0.61,
461
+ "learning_rate": 4.892082189018954e-05,
462
+ "loss": 0.2267,
463
+ "step": 365
464
+ },
465
+ {
466
+ "epoch": 0.61,
467
+ "learning_rate": 4.8881375506218976e-05,
468
+ "loss": 0.2338,
469
+ "step": 370
470
+ },
471
+ {
472
+ "epoch": 0.62,
473
+ "learning_rate": 4.8841237562267745e-05,
474
+ "loss": 0.196,
475
+ "step": 375
476
+ },
477
+ {
478
+ "epoch": 0.62,
479
+ "eval_loss": 0.16536256670951843,
480
+ "eval_runtime": 6.6441,
481
+ "eval_samples_per_second": 15.051,
482
+ "eval_steps_per_second": 15.051,
483
+ "step": 375
484
+ },
485
+ {
486
+ "epoch": 0.63,
487
+ "learning_rate": 4.880040922065566e-05,
488
+ "loss": 0.2032,
489
+ "step": 380
490
+ },
491
+ {
492
+ "epoch": 0.64,
493
+ "learning_rate": 4.875889166369517e-05,
494
+ "loss": 0.2016,
495
+ "step": 385
496
+ },
497
+ {
498
+ "epoch": 0.65,
499
+ "learning_rate": 4.8716686093657066e-05,
500
+ "loss": 0.2104,
501
+ "step": 390
502
+ },
503
+ {
504
+ "epoch": 0.66,
505
+ "learning_rate": 4.8673793732735753e-05,
506
+ "loss": 0.2049,
507
+ "step": 395
508
+ },
509
+ {
510
+ "epoch": 0.66,
511
+ "learning_rate": 4.8630215823013805e-05,
512
+ "loss": 0.2081,
513
+ "step": 400
514
+ },
515
+ {
516
+ "epoch": 0.67,
517
+ "learning_rate": 4.8585953626426e-05,
518
+ "loss": 0.2049,
519
+ "step": 405
520
+ },
521
+ {
522
+ "epoch": 0.68,
523
+ "learning_rate": 4.854100842472278e-05,
524
+ "loss": 0.2159,
525
+ "step": 410
526
+ },
527
+ {
528
+ "epoch": 0.69,
529
+ "learning_rate": 4.849538151943316e-05,
530
+ "loss": 0.2305,
531
+ "step": 415
532
+ },
533
+ {
534
+ "epoch": 0.7,
535
+ "learning_rate": 4.8449074231826994e-05,
536
+ "loss": 0.2217,
537
+ "step": 420
538
+ },
539
+ {
540
+ "epoch": 0.71,
541
+ "learning_rate": 4.840208790287676e-05,
542
+ "loss": 0.2061,
543
+ "step": 425
544
+ },
545
+ {
546
+ "epoch": 0.71,
547
+ "learning_rate": 4.835442389321867e-05,
548
+ "loss": 0.2202,
549
+ "step": 430
550
+ },
551
+ {
552
+ "epoch": 0.72,
553
+ "learning_rate": 4.8306083583113326e-05,
554
+ "loss": 0.2032,
555
+ "step": 435
556
+ },
557
+ {
558
+ "epoch": 0.73,
559
+ "learning_rate": 4.825706837240572e-05,
560
+ "loss": 0.2089,
561
+ "step": 440
562
+ },
563
+ {
564
+ "epoch": 0.74,
565
+ "learning_rate": 4.820737968048471e-05,
566
+ "loss": 0.2101,
567
+ "step": 445
568
+ },
569
+ {
570
+ "epoch": 0.75,
571
+ "learning_rate": 4.815701894624191e-05,
572
+ "loss": 0.2121,
573
+ "step": 450
574
+ },
575
+ {
576
+ "epoch": 0.76,
577
+ "learning_rate": 4.810598762803e-05,
578
+ "loss": 0.2205,
579
+ "step": 455
580
+ },
581
+ {
582
+ "epoch": 0.76,
583
+ "learning_rate": 4.805428720362054e-05,
584
+ "loss": 0.206,
585
+ "step": 460
586
+ },
587
+ {
588
+ "epoch": 0.77,
589
+ "learning_rate": 4.800191917016116e-05,
590
+ "loss": 0.2061,
591
+ "step": 465
592
+ },
593
+ {
594
+ "epoch": 0.78,
595
+ "learning_rate": 4.794888504413219e-05,
596
+ "loss": 0.2162,
597
+ "step": 470
598
+ },
599
+ {
600
+ "epoch": 0.79,
601
+ "learning_rate": 4.7895186361302734e-05,
602
+ "loss": 0.1978,
603
+ "step": 475
604
+ },
605
+ {
606
+ "epoch": 0.8,
607
+ "learning_rate": 4.784082467668627e-05,
608
+ "loss": 0.2248,
609
+ "step": 480
610
+ },
611
+ {
612
+ "epoch": 0.81,
613
+ "learning_rate": 4.7785801564495514e-05,
614
+ "loss": 0.2083,
615
+ "step": 485
616
+ },
617
+ {
618
+ "epoch": 0.81,
619
+ "learning_rate": 4.773011861809693e-05,
620
+ "loss": 0.2178,
621
+ "step": 490
622
+ },
623
+ {
624
+ "epoch": 0.82,
625
+ "learning_rate": 4.767377744996451e-05,
626
+ "loss": 0.2003,
627
+ "step": 495
628
+ },
629
+ {
630
+ "epoch": 0.83,
631
+ "learning_rate": 4.761677969163315e-05,
632
+ "loss": 0.2059,
633
+ "step": 500
634
+ },
635
+ {
636
+ "epoch": 0.83,
637
+ "eval_loss": 0.1576012670993805,
638
+ "eval_runtime": 6.6622,
639
+ "eval_samples_per_second": 15.01,
640
+ "eval_steps_per_second": 15.01,
641
+ "step": 500
642
+ },
643
+ {
644
+ "epoch": 0.84,
645
+ "learning_rate": 4.755912699365135e-05,
646
+ "loss": 0.2144,
647
+ "step": 505
648
+ },
649
+ {
650
+ "epoch": 0.85,
651
+ "learning_rate": 4.750082102553344e-05,
652
+ "loss": 0.219,
653
+ "step": 510
654
+ },
655
+ {
656
+ "epoch": 0.85,
657
+ "learning_rate": 4.7441863475711236e-05,
658
+ "loss": 0.1945,
659
+ "step": 515
660
+ },
661
+ {
662
+ "epoch": 0.86,
663
+ "learning_rate": 4.7382256051485144e-05,
664
+ "loss": 0.2298,
665
+ "step": 520
666
+ },
667
+ {
668
+ "epoch": 0.87,
669
+ "learning_rate": 4.73220004789747e-05,
670
+ "loss": 0.2052,
671
+ "step": 525
672
+ },
673
+ {
674
+ "epoch": 0.88,
675
+ "learning_rate": 4.726109850306864e-05,
676
+ "loss": 0.197,
677
+ "step": 530
678
+ },
679
+ {
680
+ "epoch": 0.89,
681
+ "learning_rate": 4.7199551887374295e-05,
682
+ "loss": 0.2122,
683
+ "step": 535
684
+ },
685
+ {
686
+ "epoch": 0.9,
687
+ "learning_rate": 4.713736241416658e-05,
688
+ "loss": 0.2088,
689
+ "step": 540
690
+ },
691
+ {
692
+ "epoch": 0.9,
693
+ "learning_rate": 4.707453188433635e-05,
694
+ "loss": 0.2086,
695
+ "step": 545
696
+ },
697
+ {
698
+ "epoch": 0.91,
699
+ "learning_rate": 4.701106211733827e-05,
700
+ "loss": 0.2022,
701
+ "step": 550
702
+ },
703
+ {
704
+ "epoch": 0.92,
705
+ "learning_rate": 4.6946954951138125e-05,
706
+ "loss": 0.2158,
707
+ "step": 555
708
+ },
709
+ {
710
+ "epoch": 0.93,
711
+ "learning_rate": 4.688221224215955e-05,
712
+ "loss": 0.2191,
713
+ "step": 560
714
+ },
715
+ {
716
+ "epoch": 0.94,
717
+ "learning_rate": 4.681683586523037e-05,
718
+ "loss": 0.2204,
719
+ "step": 565
720
+ },
721
+ {
722
+ "epoch": 0.95,
723
+ "learning_rate": 4.675082771352821e-05,
724
+ "loss": 0.2356,
725
+ "step": 570
726
+ },
727
+ {
728
+ "epoch": 0.95,
729
+ "learning_rate": 4.66841896985257e-05,
730
+ "loss": 0.2064,
731
+ "step": 575
732
+ },
733
+ {
734
+ "epoch": 0.96,
735
+ "learning_rate": 4.661692374993519e-05,
736
+ "loss": 0.2086,
737
+ "step": 580
738
+ },
739
+ {
740
+ "epoch": 0.97,
741
+ "learning_rate": 4.6549031815652744e-05,
742
+ "loss": 0.196,
743
+ "step": 585
744
+ },
745
+ {
746
+ "epoch": 0.98,
747
+ "learning_rate": 4.648051586170185e-05,
748
+ "loss": 0.2034,
749
+ "step": 590
750
+ },
751
+ {
752
+ "epoch": 0.99,
753
+ "learning_rate": 4.641137787217641e-05,
754
+ "loss": 0.2129,
755
+ "step": 595
756
+ },
757
+ {
758
+ "epoch": 1.0,
759
+ "learning_rate": 4.634161984918333e-05,
760
+ "loss": 0.2064,
761
+ "step": 600
762
+ },
763
+ {
764
+ "epoch": 1.0,
765
+ "learning_rate": 4.6271243812784514e-05,
766
+ "loss": 0.2097,
767
+ "step": 605
768
+ },
769
+ {
770
+ "epoch": 1.01,
771
+ "learning_rate": 4.62002518009384e-05,
772
+ "loss": 0.1844,
773
+ "step": 610
774
+ },
775
+ {
776
+ "epoch": 1.02,
777
+ "learning_rate": 4.612864586944089e-05,
778
+ "loss": 0.2026,
779
+ "step": 615
780
+ },
781
+ {
782
+ "epoch": 1.03,
783
+ "learning_rate": 4.605642809186589e-05,
784
+ "loss": 0.1695,
785
+ "step": 620
786
+ },
787
+ {
788
+ "epoch": 1.04,
789
+ "learning_rate": 4.598360055950519e-05,
790
+ "loss": 0.1998,
791
+ "step": 625
792
+ },
793
+ {
794
+ "epoch": 1.04,
795
+ "eval_loss": 0.15137994289398193,
796
+ "eval_runtime": 6.6769,
797
+ "eval_samples_per_second": 14.977,
798
+ "eval_steps_per_second": 14.977,
799
+ "step": 625
800
+ },
801
+ {
802
+ "epoch": 1.05,
803
+ "learning_rate": 4.591016538130796e-05,
804
+ "loss": 0.1899,
805
+ "step": 630
806
+ },
807
+ {
808
+ "epoch": 1.05,
809
+ "learning_rate": 4.583612468381964e-05,
810
+ "loss": 0.1934,
811
+ "step": 635
812
+ },
813
+ {
814
+ "epoch": 1.06,
815
+ "learning_rate": 4.576148061112039e-05,
816
+ "loss": 0.1849,
817
+ "step": 640
818
+ },
819
+ {
820
+ "epoch": 1.07,
821
+ "learning_rate": 4.5686235324762995e-05,
822
+ "loss": 0.1976,
823
+ "step": 645
824
+ },
825
+ {
826
+ "epoch": 1.08,
827
+ "learning_rate": 4.561039100371023e-05,
828
+ "loss": 0.1927,
829
+ "step": 650
830
+ },
831
+ {
832
+ "epoch": 1.09,
833
+ "learning_rate": 4.553394984427184e-05,
834
+ "loss": 0.2024,
835
+ "step": 655
836
+ },
837
+ {
838
+ "epoch": 1.1,
839
+ "learning_rate": 4.545691406004085e-05,
840
+ "loss": 0.1813,
841
+ "step": 660
842
+ },
843
+ {
844
+ "epoch": 1.1,
845
+ "learning_rate": 4.5379285881829555e-05,
846
+ "loss": 0.1706,
847
+ "step": 665
848
+ },
849
+ {
850
+ "epoch": 1.11,
851
+ "learning_rate": 4.5301067557604813e-05,
852
+ "loss": 0.1894,
853
+ "step": 670
854
+ },
855
+ {
856
+ "epoch": 1.12,
857
+ "learning_rate": 4.522226135242306e-05,
858
+ "loss": 0.1863,
859
+ "step": 675
860
+ },
861
+ {
862
+ "epoch": 1.13,
863
+ "learning_rate": 4.5142869548364644e-05,
864
+ "loss": 0.1961,
865
+ "step": 680
866
+ },
867
+ {
868
+ "epoch": 1.14,
869
+ "learning_rate": 4.5062894444467754e-05,
870
+ "loss": 0.1811,
871
+ "step": 685
872
+ },
873
+ {
874
+ "epoch": 1.15,
875
+ "learning_rate": 4.498233835666186e-05,
876
+ "loss": 0.1864,
877
+ "step": 690
878
+ },
879
+ {
880
+ "epoch": 1.15,
881
+ "learning_rate": 4.490120361770064e-05,
882
+ "loss": 0.1973,
883
+ "step": 695
884
+ },
885
+ {
886
+ "epoch": 1.16,
887
+ "learning_rate": 4.481949257709442e-05,
888
+ "loss": 0.1939,
889
+ "step": 700
890
+ },
891
+ {
892
+ "epoch": 1.17,
893
+ "learning_rate": 4.473720760104217e-05,
894
+ "loss": 0.1906,
895
+ "step": 705
896
+ },
897
+ {
898
+ "epoch": 1.18,
899
+ "learning_rate": 4.465435107236292e-05,
900
+ "loss": 0.1808,
901
+ "step": 710
902
+ },
903
+ {
904
+ "epoch": 1.19,
905
+ "learning_rate": 4.457092539042682e-05,
906
+ "loss": 0.192,
907
+ "step": 715
908
+ },
909
+ {
910
+ "epoch": 1.2,
911
+ "learning_rate": 4.448693297108563e-05,
912
+ "loss": 0.1935,
913
+ "step": 720
914
+ },
915
+ {
916
+ "epoch": 1.2,
917
+ "learning_rate": 4.440237624660275e-05,
918
+ "loss": 0.1744,
919
+ "step": 725
920
+ },
921
+ {
922
+ "epoch": 1.21,
923
+ "learning_rate": 4.4317257665582836e-05,
924
+ "loss": 0.1909,
925
+ "step": 730
926
+ },
927
+ {
928
+ "epoch": 1.22,
929
+ "learning_rate": 4.4231579692900816e-05,
930
+ "loss": 0.2009,
931
+ "step": 735
932
+ },
933
+ {
934
+ "epoch": 1.23,
935
+ "learning_rate": 4.414534480963059e-05,
936
+ "loss": 0.2023,
937
+ "step": 740
938
+ },
939
+ {
940
+ "epoch": 1.24,
941
+ "learning_rate": 4.405855551297313e-05,
942
+ "loss": 0.1932,
943
+ "step": 745
944
+ },
945
+ {
946
+ "epoch": 1.25,
947
+ "learning_rate": 4.397121431618419e-05,
948
+ "loss": 0.1798,
949
+ "step": 750
950
+ },
951
+ {
952
+ "epoch": 1.25,
953
+ "eval_loss": 0.14108647406101227,
954
+ "eval_runtime": 6.6767,
955
+ "eval_samples_per_second": 14.977,
956
+ "eval_steps_per_second": 14.977,
957
+ "step": 750
958
+ },
959
+ {
960
+ "epoch": 1.25,
961
+ "learning_rate": 4.38833237485015e-05,
962
+ "loss": 0.2114,
963
+ "step": 755
964
+ },
965
+ {
966
+ "epoch": 1.26,
967
+ "learning_rate": 4.3794886355071575e-05,
968
+ "loss": 0.1978,
969
+ "step": 760
970
+ },
971
+ {
972
+ "epoch": 1.27,
973
+ "learning_rate": 4.370590469687596e-05,
974
+ "loss": 0.1735,
975
+ "step": 765
976
+ },
977
+ {
978
+ "epoch": 1.28,
979
+ "learning_rate": 4.361638135065711e-05,
980
+ "loss": 0.1852,
981
+ "step": 770
982
+ },
983
+ {
984
+ "epoch": 1.29,
985
+ "learning_rate": 4.352631890884372e-05,
986
+ "loss": 0.1813,
987
+ "step": 775
988
+ },
989
+ {
990
+ "epoch": 1.29,
991
+ "learning_rate": 4.343571997947571e-05,
992
+ "loss": 0.1968,
993
+ "step": 780
994
+ },
995
+ {
996
+ "epoch": 1.3,
997
+ "learning_rate": 4.3344587186128676e-05,
998
+ "loss": 0.1944,
999
+ "step": 785
1000
+ },
1001
+ {
1002
+ "epoch": 1.31,
1003
+ "learning_rate": 4.325292316783791e-05,
1004
+ "loss": 0.1929,
1005
+ "step": 790
1006
+ },
1007
+ {
1008
+ "epoch": 1.32,
1009
+ "learning_rate": 4.3160730579021975e-05,
1010
+ "loss": 0.1823,
1011
+ "step": 795
1012
+ },
1013
+ {
1014
+ "epoch": 1.33,
1015
+ "learning_rate": 4.306801208940587e-05,
1016
+ "loss": 0.1968,
1017
+ "step": 800
1018
+ },
1019
+ {
1020
+ "epoch": 1.34,
1021
+ "learning_rate": 4.297477038394368e-05,
1022
+ "loss": 0.1888,
1023
+ "step": 805
1024
+ },
1025
+ {
1026
+ "epoch": 1.34,
1027
+ "learning_rate": 4.2881008162740854e-05,
1028
+ "loss": 0.196,
1029
+ "step": 810
1030
+ },
1031
+ {
1032
+ "epoch": 1.35,
1033
+ "learning_rate": 4.278672814097601e-05,
1034
+ "loss": 0.1971,
1035
+ "step": 815
1036
+ },
1037
+ {
1038
+ "epoch": 1.36,
1039
+ "learning_rate": 4.269193304882226e-05,
1040
+ "loss": 0.196,
1041
+ "step": 820
1042
+ },
1043
+ {
1044
+ "epoch": 1.37,
1045
+ "learning_rate": 4.2596625631368255e-05,
1046
+ "loss": 0.1922,
1047
+ "step": 825
1048
+ },
1049
+ {
1050
+ "epoch": 1.38,
1051
+ "learning_rate": 4.250080864853858e-05,
1052
+ "loss": 0.1955,
1053
+ "step": 830
1054
+ },
1055
+ {
1056
+ "epoch": 1.39,
1057
+ "learning_rate": 4.2404484875013895e-05,
1058
+ "loss": 0.1892,
1059
+ "step": 835
1060
+ },
1061
+ {
1062
+ "epoch": 1.39,
1063
+ "learning_rate": 4.2307657100150584e-05,
1064
+ "loss": 0.1903,
1065
+ "step": 840
1066
+ },
1067
+ {
1068
+ "epoch": 1.4,
1069
+ "learning_rate": 4.2210328127899943e-05,
1070
+ "loss": 0.1953,
1071
+ "step": 845
1072
+ },
1073
+ {
1074
+ "epoch": 1.41,
1075
+ "learning_rate": 4.211250077672704e-05,
1076
+ "loss": 0.1916,
1077
+ "step": 850
1078
+ },
1079
+ {
1080
+ "epoch": 1.42,
1081
+ "learning_rate": 4.201417787952901e-05,
1082
+ "loss": 0.1951,
1083
+ "step": 855
1084
+ },
1085
+ {
1086
+ "epoch": 1.43,
1087
+ "learning_rate": 4.191536228355315e-05,
1088
+ "loss": 0.181,
1089
+ "step": 860
1090
+ },
1091
+ {
1092
+ "epoch": 1.44,
1093
+ "learning_rate": 4.181605685031432e-05,
1094
+ "loss": 0.1976,
1095
+ "step": 865
1096
+ },
1097
+ {
1098
+ "epoch": 1.44,
1099
+ "learning_rate": 4.17162644555122e-05,
1100
+ "loss": 0.1915,
1101
+ "step": 870
1102
+ },
1103
+ {
1104
+ "epoch": 1.45,
1105
+ "learning_rate": 4.1615987988947944e-05,
1106
+ "loss": 0.2061,
1107
+ "step": 875
1108
+ },
1109
+ {
1110
+ "epoch": 1.45,
1111
+ "eval_loss": 0.13502272963523865,
1112
+ "eval_runtime": 6.5739,
1113
+ "eval_samples_per_second": 15.212,
1114
+ "eval_steps_per_second": 15.212,
1115
+ "step": 875
1116
+ },
1117
+ {
1118
+ "epoch": 1.46,
1119
+ "learning_rate": 4.1515230354440524e-05,
1120
+ "loss": 0.2027,
1121
+ "step": 880
1122
+ },
1123
+ {
1124
+ "epoch": 1.47,
1125
+ "learning_rate": 4.1413994469742637e-05,
1126
+ "loss": 0.1916,
1127
+ "step": 885
1128
+ },
1129
+ {
1130
+ "epoch": 1.48,
1131
+ "learning_rate": 4.131228326645621e-05,
1132
+ "loss": 0.1925,
1133
+ "step": 890
1134
+ },
1135
+ {
1136
+ "epoch": 1.49,
1137
+ "learning_rate": 4.1210099689947513e-05,
1138
+ "loss": 0.1941,
1139
+ "step": 895
1140
+ },
1141
+ {
1142
+ "epoch": 1.49,
1143
+ "learning_rate": 4.110744669926184e-05,
1144
+ "loss": 0.1746,
1145
+ "step": 900
1146
+ },
1147
+ {
1148
+ "epoch": 1.5,
1149
+ "learning_rate": 4.100432726703788e-05,
1150
+ "loss": 0.1876,
1151
+ "step": 905
1152
+ },
1153
+ {
1154
+ "epoch": 1.51,
1155
+ "learning_rate": 4.090074437942155e-05,
1156
+ "loss": 0.204,
1157
+ "step": 910
1158
+ },
1159
+ {
1160
+ "epoch": 1.52,
1161
+ "learning_rate": 4.0796701035979626e-05,
1162
+ "loss": 0.1852,
1163
+ "step": 915
1164
+ },
1165
+ {
1166
+ "epoch": 1.53,
1167
+ "learning_rate": 4.069220024961275e-05,
1168
+ "loss": 0.1925,
1169
+ "step": 920
1170
+ },
1171
+ {
1172
+ "epoch": 1.54,
1173
+ "learning_rate": 4.058724504646834e-05,
1174
+ "loss": 0.192,
1175
+ "step": 925
1176
+ },
1177
+ {
1178
+ "epoch": 1.54,
1179
+ "learning_rate": 4.0481838465852806e-05,
1180
+ "loss": 0.2025,
1181
+ "step": 930
1182
+ },
1183
+ {
1184
+ "epoch": 1.55,
1185
+ "learning_rate": 4.037598356014363e-05,
1186
+ "loss": 0.1921,
1187
+ "step": 935
1188
+ },
1189
+ {
1190
+ "epoch": 1.56,
1191
+ "learning_rate": 4.026968339470096e-05,
1192
+ "loss": 0.1824,
1193
+ "step": 940
1194
+ },
1195
+ {
1196
+ "epoch": 1.57,
1197
+ "learning_rate": 4.0162941047778825e-05,
1198
+ "loss": 0.1861,
1199
+ "step": 945
1200
+ },
1201
+ {
1202
+ "epoch": 1.58,
1203
+ "learning_rate": 4.0055759610435996e-05,
1204
+ "loss": 0.1928,
1205
+ "step": 950
1206
+ },
1207
+ {
1208
+ "epoch": 1.59,
1209
+ "learning_rate": 3.9948142186446486e-05,
1210
+ "loss": 0.1821,
1211
+ "step": 955
1212
+ },
1213
+ {
1214
+ "epoch": 1.59,
1215
+ "learning_rate": 3.9840091892209673e-05,
1216
+ "loss": 0.1918,
1217
+ "step": 960
1218
+ },
1219
+ {
1220
+ "epoch": 1.6,
1221
+ "learning_rate": 3.973161185666005e-05,
1222
+ "loss": 0.1779,
1223
+ "step": 965
1224
+ },
1225
+ {
1226
+ "epoch": 1.61,
1227
+ "learning_rate": 3.9622705221176596e-05,
1228
+ "loss": 0.1854,
1229
+ "step": 970
1230
+ },
1231
+ {
1232
+ "epoch": 1.62,
1233
+ "learning_rate": 3.951337513949184e-05,
1234
+ "loss": 0.1732,
1235
+ "step": 975
1236
+ },
1237
+ {
1238
+ "epoch": 1.63,
1239
+ "learning_rate": 3.940362477760052e-05,
1240
+ "loss": 0.1893,
1241
+ "step": 980
1242
+ },
1243
+ {
1244
+ "epoch": 1.64,
1245
+ "learning_rate": 3.92934573136679e-05,
1246
+ "loss": 0.1969,
1247
+ "step": 985
1248
+ },
1249
+ {
1250
+ "epoch": 1.64,
1251
+ "learning_rate": 3.9182875937937755e-05,
1252
+ "loss": 0.1884,
1253
+ "step": 990
1254
+ },
1255
+ {
1256
+ "epoch": 1.65,
1257
+ "learning_rate": 3.907188385263995e-05,
1258
+ "loss": 0.1837,
1259
+ "step": 995
1260
+ },
1261
+ {
1262
+ "epoch": 1.66,
1263
+ "learning_rate": 3.8960484271897736e-05,
1264
+ "loss": 0.1887,
1265
+ "step": 1000
1266
+ },
1267
+ {
1268
+ "epoch": 1.66,
1269
+ "eval_loss": 0.13266006112098694,
1270
+ "eval_runtime": 6.6053,
1271
+ "eval_samples_per_second": 15.139,
1272
+ "eval_steps_per_second": 15.139,
1273
+ "step": 1000
1274
+ },
1275
+ {
1276
+ "epoch": 1.67,
1277
+ "learning_rate": 3.8848680421634676e-05,
1278
+ "loss": 0.1837,
1279
+ "step": 1005
1280
+ },
1281
+ {
1282
+ "epoch": 1.68,
1283
+ "learning_rate": 3.873647553948121e-05,
1284
+ "loss": 0.199,
1285
+ "step": 1010
1286
+ },
1287
+ {
1288
+ "epoch": 1.68,
1289
+ "learning_rate": 3.862387287468094e-05,
1290
+ "loss": 0.1933,
1291
+ "step": 1015
1292
+ },
1293
+ {
1294
+ "epoch": 1.69,
1295
+ "learning_rate": 3.851087568799648e-05,
1296
+ "loss": 0.1761,
1297
+ "step": 1020
1298
+ },
1299
+ {
1300
+ "epoch": 1.7,
1301
+ "learning_rate": 3.8397487251615085e-05,
1302
+ "loss": 0.1848,
1303
+ "step": 1025
1304
+ },
1305
+ {
1306
+ "epoch": 1.71,
1307
+ "learning_rate": 3.8283710849053834e-05,
1308
+ "loss": 0.1883,
1309
+ "step": 1030
1310
+ },
1311
+ {
1312
+ "epoch": 1.72,
1313
+ "learning_rate": 3.816954977506462e-05,
1314
+ "loss": 0.2009,
1315
+ "step": 1035
1316
+ },
1317
+ {
1318
+ "epoch": 1.73,
1319
+ "learning_rate": 3.8055007335538665e-05,
1320
+ "loss": 0.1863,
1321
+ "step": 1040
1322
+ },
1323
+ {
1324
+ "epoch": 1.73,
1325
+ "learning_rate": 3.7940086847410874e-05,
1326
+ "loss": 0.1866,
1327
+ "step": 1045
1328
+ },
1329
+ {
1330
+ "epoch": 1.74,
1331
+ "learning_rate": 3.782479163856367e-05,
1332
+ "loss": 0.2085,
1333
+ "step": 1050
1334
+ },
1335
+ {
1336
+ "epoch": 1.75,
1337
+ "learning_rate": 3.770912504773076e-05,
1338
+ "loss": 0.1874,
1339
+ "step": 1055
1340
+ },
1341
+ {
1342
+ "epoch": 1.76,
1343
+ "learning_rate": 3.759309042440028e-05,
1344
+ "loss": 0.1885,
1345
+ "step": 1060
1346
+ },
1347
+ {
1348
+ "epoch": 1.77,
1349
+ "learning_rate": 3.747669112871801e-05,
1350
+ "loss": 0.195,
1351
+ "step": 1065
1352
+ },
1353
+ {
1354
+ "epoch": 1.78,
1355
+ "learning_rate": 3.735993053138988e-05,
1356
+ "loss": 0.1969,
1357
+ "step": 1070
1358
+ },
1359
+ {
1360
+ "epoch": 1.78,
1361
+ "learning_rate": 3.724281201358447e-05,
1362
+ "loss": 0.193,
1363
+ "step": 1075
1364
+ },
1365
+ {
1366
+ "epoch": 1.79,
1367
+ "learning_rate": 3.7125338966835064e-05,
1368
+ "loss": 0.179,
1369
+ "step": 1080
1370
+ },
1371
+ {
1372
+ "epoch": 1.8,
1373
+ "learning_rate": 3.700751479294146e-05,
1374
+ "loss": 0.2007,
1375
+ "step": 1085
1376
+ },
1377
+ {
1378
+ "epoch": 1.81,
1379
+ "learning_rate": 3.68893429038714e-05,
1380
+ "loss": 0.1979,
1381
+ "step": 1090
1382
+ },
1383
+ {
1384
+ "epoch": 1.82,
1385
+ "learning_rate": 3.677082672166185e-05,
1386
+ "loss": 0.1982,
1387
+ "step": 1095
1388
+ },
1389
+ {
1390
+ "epoch": 1.83,
1391
+ "learning_rate": 3.665196967831981e-05,
1392
+ "loss": 0.1936,
1393
+ "step": 1100
1394
+ },
1395
+ {
1396
+ "epoch": 1.83,
1397
+ "learning_rate": 3.6532775215723044e-05,
1398
+ "loss": 0.1896,
1399
+ "step": 1105
1400
+ },
1401
+ {
1402
+ "epoch": 1.84,
1403
+ "learning_rate": 3.641324678552031e-05,
1404
+ "loss": 0.1927,
1405
+ "step": 1110
1406
+ },
1407
+ {
1408
+ "epoch": 1.85,
1409
+ "learning_rate": 3.629338784903145e-05,
1410
+ "loss": 0.1961,
1411
+ "step": 1115
1412
+ },
1413
+ {
1414
+ "epoch": 1.86,
1415
+ "learning_rate": 3.6173201877147134e-05,
1416
+ "loss": 0.2042,
1417
+ "step": 1120
1418
+ },
1419
+ {
1420
+ "epoch": 1.87,
1421
+ "learning_rate": 3.6052692350228404e-05,
1422
+ "loss": 0.1966,
1423
+ "step": 1125
1424
+ },
1425
+ {
1426
+ "epoch": 1.87,
1427
+ "eval_loss": 0.12441341578960419,
1428
+ "eval_runtime": 6.5924,
1429
+ "eval_samples_per_second": 15.169,
1430
+ "eval_steps_per_second": 15.169,
1431
+ "step": 1125
1432
+ },
1433
+ {
1434
+ "epoch": 1.88,
1435
+ "learning_rate": 3.5931862758005803e-05,
1436
+ "loss": 0.1934,
1437
+ "step": 1130
1438
+ },
1439
+ {
1440
+ "epoch": 1.88,
1441
+ "learning_rate": 3.581071659947842e-05,
1442
+ "loss": 0.1776,
1443
+ "step": 1135
1444
+ },
1445
+ {
1446
+ "epoch": 1.89,
1447
+ "learning_rate": 3.5689257382812454e-05,
1448
+ "loss": 0.1814,
1449
+ "step": 1140
1450
+ },
1451
+ {
1452
+ "epoch": 1.9,
1453
+ "learning_rate": 3.556748862523973e-05,
1454
+ "loss": 0.1909,
1455
+ "step": 1145
1456
+ },
1457
+ {
1458
+ "epoch": 1.91,
1459
+ "learning_rate": 3.544541385295578e-05,
1460
+ "loss": 0.1925,
1461
+ "step": 1150
1462
+ },
1463
+ {
1464
+ "epoch": 1.92,
1465
+ "learning_rate": 3.5323036601017756e-05,
1466
+ "loss": 0.1847,
1467
+ "step": 1155
1468
+ },
1469
+ {
1470
+ "epoch": 1.93,
1471
+ "learning_rate": 3.520036041324203e-05,
1472
+ "loss": 0.1779,
1473
+ "step": 1160
1474
+ },
1475
+ {
1476
+ "epoch": 1.93,
1477
+ "learning_rate": 3.507738884210164e-05,
1478
+ "loss": 0.1907,
1479
+ "step": 1165
1480
+ },
1481
+ {
1482
+ "epoch": 1.94,
1483
+ "learning_rate": 3.4954125448623336e-05,
1484
+ "loss": 0.1807,
1485
+ "step": 1170
1486
+ },
1487
+ {
1488
+ "epoch": 1.95,
1489
+ "learning_rate": 3.483057380228452e-05,
1490
+ "loss": 0.1777,
1491
+ "step": 1175
1492
+ },
1493
+ {
1494
+ "epoch": 1.96,
1495
+ "learning_rate": 3.470673748090984e-05,
1496
+ "loss": 0.194,
1497
+ "step": 1180
1498
+ },
1499
+ {
1500
+ "epoch": 1.97,
1501
+ "learning_rate": 3.4582620070567614e-05,
1502
+ "loss": 0.1821,
1503
+ "step": 1185
1504
+ },
1505
+ {
1506
+ "epoch": 1.98,
1507
+ "learning_rate": 3.445822516546598e-05,
1508
+ "loss": 0.1915,
1509
+ "step": 1190
1510
+ },
1511
+ {
1512
+ "epoch": 1.98,
1513
+ "learning_rate": 3.4333556367848776e-05,
1514
+ "loss": 0.1688,
1515
+ "step": 1195
1516
+ },
1517
+ {
1518
+ "epoch": 1.99,
1519
+ "learning_rate": 3.420861728789129e-05,
1520
+ "loss": 0.1949,
1521
+ "step": 1200
1522
+ },
1523
+ {
1524
+ "epoch": 2.0,
1525
+ "learning_rate": 3.408341154359564e-05,
1526
+ "loss": 0.1974,
1527
+ "step": 1205
1528
+ },
1529
+ {
1530
+ "epoch": 2.01,
1531
+ "learning_rate": 3.395794276068609e-05,
1532
+ "loss": 0.1609,
1533
+ "step": 1210
1534
+ },
1535
+ {
1536
+ "epoch": 2.02,
1537
+ "learning_rate": 3.3832214572503975e-05,
1538
+ "loss": 0.1667,
1539
+ "step": 1215
1540
+ },
1541
+ {
1542
+ "epoch": 2.03,
1543
+ "learning_rate": 3.370623061990255e-05,
1544
+ "loss": 0.1726,
1545
+ "step": 1220
1546
+ },
1547
+ {
1548
+ "epoch": 2.03,
1549
+ "learning_rate": 3.357999455114148e-05,
1550
+ "loss": 0.1659,
1551
+ "step": 1225
1552
+ },
1553
+ {
1554
+ "epoch": 2.04,
1555
+ "learning_rate": 3.3453510021781325e-05,
1556
+ "loss": 0.1694,
1557
+ "step": 1230
1558
+ },
1559
+ {
1560
+ "epoch": 2.05,
1561
+ "learning_rate": 3.3326780694577506e-05,
1562
+ "loss": 0.1668,
1563
+ "step": 1235
1564
+ },
1565
+ {
1566
+ "epoch": 2.06,
1567
+ "learning_rate": 3.319981023937442e-05,
1568
+ "loss": 0.1715,
1569
+ "step": 1240
1570
+ },
1571
+ {
1572
+ "epoch": 2.07,
1573
+ "learning_rate": 3.3072602332999e-05,
1574
+ "loss": 0.1698,
1575
+ "step": 1245
1576
+ },
1577
+ {
1578
+ "epoch": 2.08,
1579
+ "learning_rate": 3.294516065915438e-05,
1580
+ "loss": 0.174,
1581
+ "step": 1250
1582
+ },
1583
+ {
1584
+ "epoch": 2.08,
1585
+ "eval_loss": 0.11452513933181763,
1586
+ "eval_runtime": 6.5986,
1587
+ "eval_samples_per_second": 15.155,
1588
+ "eval_steps_per_second": 15.155,
1589
+ "step": 1250
1590
+ },
1591
+ {
1592
+ "epoch": 2.08,
1593
+ "learning_rate": 3.2817488908313134e-05,
1594
+ "loss": 0.1546,
1595
+ "step": 1255
1596
+ },
1597
+ {
1598
+ "epoch": 2.09,
1599
+ "learning_rate": 3.268959077761044e-05,
1600
+ "loss": 0.1739,
1601
+ "step": 1260
1602
+ },
1603
+ {
1604
+ "epoch": 2.1,
1605
+ "learning_rate": 3.256146997073702e-05,
1606
+ "loss": 0.1765,
1607
+ "step": 1265
1608
+ },
1609
+ {
1610
+ "epoch": 2.11,
1611
+ "learning_rate": 3.243313019783187e-05,
1612
+ "loss": 0.174,
1613
+ "step": 1270
1614
+ },
1615
+ {
1616
+ "epoch": 2.12,
1617
+ "learning_rate": 3.230457517537485e-05,
1618
+ "loss": 0.1673,
1619
+ "step": 1275
1620
+ },
1621
+ {
1622
+ "epoch": 2.12,
1623
+ "learning_rate": 3.217580862607904e-05,
1624
+ "loss": 0.1682,
1625
+ "step": 1280
1626
+ },
1627
+ {
1628
+ "epoch": 2.13,
1629
+ "learning_rate": 3.204683427878291e-05,
1630
+ "loss": 0.1641,
1631
+ "step": 1285
1632
+ },
1633
+ {
1634
+ "epoch": 2.14,
1635
+ "learning_rate": 3.191765586834243e-05,
1636
+ "loss": 0.1763,
1637
+ "step": 1290
1638
+ },
1639
+ {
1640
+ "epoch": 2.15,
1641
+ "learning_rate": 3.178827713552281e-05,
1642
+ "loss": 0.1628,
1643
+ "step": 1295
1644
+ },
1645
+ {
1646
+ "epoch": 2.16,
1647
+ "learning_rate": 3.165870182689024e-05,
1648
+ "loss": 0.1648,
1649
+ "step": 1300
1650
+ },
1651
+ {
1652
+ "epoch": 2.17,
1653
+ "learning_rate": 3.152893369470337e-05,
1654
+ "loss": 0.1773,
1655
+ "step": 1305
1656
+ },
1657
+ {
1658
+ "epoch": 2.17,
1659
+ "learning_rate": 3.139897649680466e-05,
1660
+ "loss": 0.1672,
1661
+ "step": 1310
1662
+ },
1663
+ {
1664
+ "epoch": 2.18,
1665
+ "learning_rate": 3.126883399651158e-05,
1666
+ "loss": 0.1633,
1667
+ "step": 1315
1668
+ },
1669
+ {
1670
+ "epoch": 2.19,
1671
+ "learning_rate": 3.1138509962507585e-05,
1672
+ "loss": 0.1543,
1673
+ "step": 1320
1674
+ },
1675
+ {
1676
+ "epoch": 2.2,
1677
+ "learning_rate": 3.100800816873301e-05,
1678
+ "loss": 0.1667,
1679
+ "step": 1325
1680
+ },
1681
+ {
1682
+ "epoch": 2.21,
1683
+ "learning_rate": 3.08773323942758e-05,
1684
+ "loss": 0.1568,
1685
+ "step": 1330
1686
+ },
1687
+ {
1688
+ "epoch": 2.22,
1689
+ "learning_rate": 3.074648642326204e-05,
1690
+ "loss": 0.1622,
1691
+ "step": 1335
1692
+ },
1693
+ {
1694
+ "epoch": 2.22,
1695
+ "learning_rate": 3.0615474044746386e-05,
1696
+ "loss": 0.1609,
1697
+ "step": 1340
1698
+ },
1699
+ {
1700
+ "epoch": 2.23,
1701
+ "learning_rate": 3.0484299052602355e-05,
1702
+ "loss": 0.1647,
1703
+ "step": 1345
1704
+ },
1705
+ {
1706
+ "epoch": 2.24,
1707
+ "learning_rate": 3.0352965245412445e-05,
1708
+ "loss": 0.1775,
1709
+ "step": 1350
1710
+ },
1711
+ {
1712
+ "epoch": 2.25,
1713
+ "learning_rate": 3.022147642635813e-05,
1714
+ "loss": 0.1762,
1715
+ "step": 1355
1716
+ },
1717
+ {
1718
+ "epoch": 2.26,
1719
+ "learning_rate": 3.0089836403109762e-05,
1720
+ "loss": 0.1603,
1721
+ "step": 1360
1722
+ },
1723
+ {
1724
+ "epoch": 2.27,
1725
+ "learning_rate": 2.9958048987716265e-05,
1726
+ "loss": 0.1581,
1727
+ "step": 1365
1728
+ },
1729
+ {
1730
+ "epoch": 2.27,
1731
+ "learning_rate": 2.9826117996494762e-05,
1732
+ "loss": 0.1643,
1733
+ "step": 1370
1734
+ },
1735
+ {
1736
+ "epoch": 2.28,
1737
+ "learning_rate": 2.9694047249920092e-05,
1738
+ "loss": 0.1722,
1739
+ "step": 1375
1740
+ },
1741
+ {
1742
+ "epoch": 2.28,
1743
+ "eval_loss": 0.10874178260564804,
1744
+ "eval_runtime": 6.6047,
1745
+ "eval_samples_per_second": 15.141,
1746
+ "eval_steps_per_second": 15.141,
1747
+ "step": 1375
1748
+ },
1749
+ {
1750
+ "epoch": 2.29,
1751
+ "learning_rate": 2.95618405725141e-05,
1752
+ "loss": 0.1739,
1753
+ "step": 1380
1754
+ },
1755
+ {
1756
+ "epoch": 2.3,
1757
+ "learning_rate": 2.9429501792734987e-05,
1758
+ "loss": 0.1703,
1759
+ "step": 1385
1760
+ },
1761
+ {
1762
+ "epoch": 2.31,
1763
+ "learning_rate": 2.9297034742866337e-05,
1764
+ "loss": 0.1615,
1765
+ "step": 1390
1766
+ },
1767
+ {
1768
+ "epoch": 2.32,
1769
+ "learning_rate": 2.9164443258906245e-05,
1770
+ "loss": 0.1658,
1771
+ "step": 1395
1772
+ },
1773
+ {
1774
+ "epoch": 2.32,
1775
+ "learning_rate": 2.9031731180456158e-05,
1776
+ "loss": 0.159,
1777
+ "step": 1400
1778
+ },
1779
+ {
1780
+ "epoch": 2.33,
1781
+ "learning_rate": 2.889890235060972e-05,
1782
+ "loss": 0.1565,
1783
+ "step": 1405
1784
+ },
1785
+ {
1786
+ "epoch": 2.34,
1787
+ "learning_rate": 2.876596061584147e-05,
1788
+ "loss": 0.163,
1789
+ "step": 1410
1790
+ },
1791
+ {
1792
+ "epoch": 2.35,
1793
+ "learning_rate": 2.8632909825895477e-05,
1794
+ "loss": 0.1737,
1795
+ "step": 1415
1796
+ },
1797
+ {
1798
+ "epoch": 2.36,
1799
+ "learning_rate": 2.8499753833673824e-05,
1800
+ "loss": 0.1633,
1801
+ "step": 1420
1802
+ },
1803
+ {
1804
+ "epoch": 2.37,
1805
+ "learning_rate": 2.8366496495125096e-05,
1806
+ "loss": 0.1697,
1807
+ "step": 1425
1808
+ },
1809
+ {
1810
+ "epoch": 2.37,
1811
+ "learning_rate": 2.8233141669132607e-05,
1812
+ "loss": 0.1669,
1813
+ "step": 1430
1814
+ },
1815
+ {
1816
+ "epoch": 2.38,
1817
+ "learning_rate": 2.8099693217402806e-05,
1818
+ "loss": 0.1601,
1819
+ "step": 1435
1820
+ },
1821
+ {
1822
+ "epoch": 2.39,
1823
+ "learning_rate": 2.7966155004353306e-05,
1824
+ "loss": 0.1641,
1825
+ "step": 1440
1826
+ },
1827
+ {
1828
+ "epoch": 2.4,
1829
+ "learning_rate": 2.783253089700108e-05,
1830
+ "loss": 0.1929,
1831
+ "step": 1445
1832
+ },
1833
+ {
1834
+ "epoch": 2.41,
1835
+ "learning_rate": 2.7698824764850408e-05,
1836
+ "loss": 0.1745,
1837
+ "step": 1450
1838
+ },
1839
+ {
1840
+ "epoch": 2.42,
1841
+ "learning_rate": 2.756504047978088e-05,
1842
+ "loss": 0.165,
1843
+ "step": 1455
1844
+ },
1845
+ {
1846
+ "epoch": 2.42,
1847
+ "learning_rate": 2.743118191593523e-05,
1848
+ "loss": 0.1598,
1849
+ "step": 1460
1850
+ },
1851
+ {
1852
+ "epoch": 2.43,
1853
+ "learning_rate": 2.7297252949607193e-05,
1854
+ "loss": 0.1579,
1855
+ "step": 1465
1856
+ },
1857
+ {
1858
+ "epoch": 2.44,
1859
+ "learning_rate": 2.716325745912918e-05,
1860
+ "loss": 0.1681,
1861
+ "step": 1470
1862
+ },
1863
+ {
1864
+ "epoch": 2.45,
1865
+ "learning_rate": 2.7029199324760078e-05,
1866
+ "loss": 0.1763,
1867
+ "step": 1475
1868
+ },
1869
+ {
1870
+ "epoch": 2.46,
1871
+ "learning_rate": 2.6895082428572776e-05,
1872
+ "loss": 0.1653,
1873
+ "step": 1480
1874
+ },
1875
+ {
1876
+ "epoch": 2.47,
1877
+ "learning_rate": 2.6760910654341816e-05,
1878
+ "loss": 0.1564,
1879
+ "step": 1485
1880
+ },
1881
+ {
1882
+ "epoch": 2.47,
1883
+ "learning_rate": 2.66266878874309e-05,
1884
+ "loss": 0.1587,
1885
+ "step": 1490
1886
+ },
1887
+ {
1888
+ "epoch": 2.48,
1889
+ "learning_rate": 2.6492418014680385e-05,
1890
+ "loss": 0.1532,
1891
+ "step": 1495
1892
+ },
1893
+ {
1894
+ "epoch": 2.49,
1895
+ "learning_rate": 2.635810492429473e-05,
1896
+ "loss": 0.1717,
1897
+ "step": 1500
1898
+ },
1899
+ {
1900
+ "epoch": 2.49,
1901
+ "eval_loss": 0.10279469192028046,
1902
+ "eval_runtime": 6.5822,
1903
+ "eval_samples_per_second": 15.193,
1904
+ "eval_steps_per_second": 15.193,
1905
+ "step": 1500
1906
+ },
1907
+ {
1908
+ "epoch": 2.5,
1909
+ "learning_rate": 2.622375250572988e-05,
1910
+ "loss": 0.1849,
1911
+ "step": 1505
1912
+ },
1913
+ {
1914
+ "epoch": 2.51,
1915
+ "learning_rate": 2.6089364649580676e-05,
1916
+ "loss": 0.1573,
1917
+ "step": 1510
1918
+ },
1919
+ {
1920
+ "epoch": 2.52,
1921
+ "learning_rate": 2.5954945247468137e-05,
1922
+ "loss": 0.1748,
1923
+ "step": 1515
1924
+ },
1925
+ {
1926
+ "epoch": 2.52,
1927
+ "learning_rate": 2.5820498191926816e-05,
1928
+ "loss": 0.1663,
1929
+ "step": 1520
1930
+ },
1931
+ {
1932
+ "epoch": 2.53,
1933
+ "learning_rate": 2.5686027376292037e-05,
1934
+ "loss": 0.1678,
1935
+ "step": 1525
1936
+ },
1937
+ {
1938
+ "epoch": 2.54,
1939
+ "learning_rate": 2.5551536694587203e-05,
1940
+ "loss": 0.1633,
1941
+ "step": 1530
1942
+ },
1943
+ {
1944
+ "epoch": 2.55,
1945
+ "learning_rate": 2.541703004141096e-05,
1946
+ "loss": 0.1543,
1947
+ "step": 1535
1948
+ },
1949
+ {
1950
+ "epoch": 2.56,
1951
+ "learning_rate": 2.5282511311824497e-05,
1952
+ "loss": 0.1719,
1953
+ "step": 1540
1954
+ },
1955
+ {
1956
+ "epoch": 2.56,
1957
+ "learning_rate": 2.514798440123869e-05,
1958
+ "loss": 0.1532,
1959
+ "step": 1545
1960
+ },
1961
+ {
1962
+ "epoch": 2.57,
1963
+ "learning_rate": 2.501345320530133e-05,
1964
+ "loss": 0.1539,
1965
+ "step": 1550
1966
+ },
1967
+ {
1968
+ "epoch": 2.58,
1969
+ "learning_rate": 2.4878921619784295e-05,
1970
+ "loss": 0.1671,
1971
+ "step": 1555
1972
+ },
1973
+ {
1974
+ "epoch": 2.59,
1975
+ "learning_rate": 2.474439354047076e-05,
1976
+ "loss": 0.1702,
1977
+ "step": 1560
1978
+ },
1979
+ {
1980
+ "epoch": 2.6,
1981
+ "learning_rate": 2.4609872863042344e-05,
1982
+ "loss": 0.1835,
1983
+ "step": 1565
1984
+ },
1985
+ {
1986
+ "epoch": 2.61,
1987
+ "learning_rate": 2.4475363482966353e-05,
1988
+ "loss": 0.1621,
1989
+ "step": 1570
1990
+ },
1991
+ {
1992
+ "epoch": 2.61,
1993
+ "learning_rate": 2.4340869295382924e-05,
1994
+ "loss": 0.1673,
1995
+ "step": 1575
1996
+ },
1997
+ {
1998
+ "epoch": 2.62,
1999
+ "learning_rate": 2.4206394194992232e-05,
2000
+ "loss": 0.1648,
2001
+ "step": 1580
2002
+ },
2003
+ {
2004
+ "epoch": 2.63,
2005
+ "learning_rate": 2.4071942075941746e-05,
2006
+ "loss": 0.162,
2007
+ "step": 1585
2008
+ },
2009
+ {
2010
+ "epoch": 2.64,
2011
+ "learning_rate": 2.3937516831713438e-05,
2012
+ "loss": 0.1881,
2013
+ "step": 1590
2014
+ },
2015
+ {
2016
+ "epoch": 2.65,
2017
+ "learning_rate": 2.380312235501102e-05,
2018
+ "loss": 0.1662,
2019
+ "step": 1595
2020
+ },
2021
+ {
2022
+ "epoch": 2.66,
2023
+ "learning_rate": 2.3668762537647254e-05,
2024
+ "loss": 0.1641,
2025
+ "step": 1600
2026
+ },
2027
+ {
2028
+ "epoch": 2.66,
2029
+ "learning_rate": 2.3534441270431225e-05,
2030
+ "loss": 0.16,
2031
+ "step": 1605
2032
+ },
2033
+ {
2034
+ "epoch": 2.67,
2035
+ "learning_rate": 2.3400162443055656e-05,
2036
+ "loss": 0.1682,
2037
+ "step": 1610
2038
+ },
2039
+ {
2040
+ "epoch": 2.68,
2041
+ "learning_rate": 2.3265929943984314e-05,
2042
+ "loss": 0.1821,
2043
+ "step": 1615
2044
+ },
2045
+ {
2046
+ "epoch": 2.69,
2047
+ "learning_rate": 2.3131747660339394e-05,
2048
+ "loss": 0.1728,
2049
+ "step": 1620
2050
+ },
2051
+ {
2052
+ "epoch": 2.7,
2053
+ "learning_rate": 2.2997619477788924e-05,
2054
+ "loss": 0.1596,
2055
+ "step": 1625
2056
+ },
2057
+ {
2058
+ "epoch": 2.7,
2059
+ "eval_loss": 0.0988275557756424,
2060
+ "eval_runtime": 6.5679,
2061
+ "eval_samples_per_second": 15.226,
2062
+ "eval_steps_per_second": 15.226,
2063
+ "step": 1625
2064
+ },
2065
+ {
2066
+ "epoch": 2.71,
2067
+ "learning_rate": 2.2863549280434282e-05,
2068
+ "loss": 0.1636,
2069
+ "step": 1630
2070
+ },
2071
+ {
2072
+ "epoch": 2.71,
2073
+ "learning_rate": 2.2729540950697677e-05,
2074
+ "loss": 0.1544,
2075
+ "step": 1635
2076
+ },
2077
+ {
2078
+ "epoch": 2.72,
2079
+ "learning_rate": 2.2595598369209782e-05,
2080
+ "loss": 0.1867,
2081
+ "step": 1640
2082
+ },
2083
+ {
2084
+ "epoch": 2.73,
2085
+ "learning_rate": 2.2461725414697303e-05,
2086
+ "loss": 0.1592,
2087
+ "step": 1645
2088
+ },
2089
+ {
2090
+ "epoch": 2.74,
2091
+ "learning_rate": 2.232792596387068e-05,
2092
+ "loss": 0.1583,
2093
+ "step": 1650
2094
+ },
2095
+ {
2096
+ "epoch": 2.75,
2097
+ "learning_rate": 2.2194203891311842e-05,
2098
+ "loss": 0.1569,
2099
+ "step": 1655
2100
+ },
2101
+ {
2102
+ "epoch": 2.76,
2103
+ "learning_rate": 2.2060563069361955e-05,
2104
+ "loss": 0.1698,
2105
+ "step": 1660
2106
+ },
2107
+ {
2108
+ "epoch": 2.76,
2109
+ "learning_rate": 2.1927007368009345e-05,
2110
+ "loss": 0.1771,
2111
+ "step": 1665
2112
+ },
2113
+ {
2114
+ "epoch": 2.77,
2115
+ "learning_rate": 2.17935406547774e-05,
2116
+ "loss": 0.1644,
2117
+ "step": 1670
2118
+ },
2119
+ {
2120
+ "epoch": 2.78,
2121
+ "learning_rate": 2.1660166794612575e-05,
2122
+ "loss": 0.1586,
2123
+ "step": 1675
2124
+ },
2125
+ {
2126
+ "epoch": 2.79,
2127
+ "learning_rate": 2.1526889649772478e-05,
2128
+ "loss": 0.172,
2129
+ "step": 1680
2130
+ },
2131
+ {
2132
+ "epoch": 2.8,
2133
+ "learning_rate": 2.1393713079714e-05,
2134
+ "loss": 0.1558,
2135
+ "step": 1685
2136
+ },
2137
+ {
2138
+ "epoch": 2.81,
2139
+ "learning_rate": 2.1260640940981617e-05,
2140
+ "loss": 0.168,
2141
+ "step": 1690
2142
+ },
2143
+ {
2144
+ "epoch": 2.81,
2145
+ "learning_rate": 2.1127677087095632e-05,
2146
+ "loss": 0.1763,
2147
+ "step": 1695
2148
+ },
2149
+ {
2150
+ "epoch": 2.82,
2151
+ "learning_rate": 2.099482536844063e-05,
2152
+ "loss": 0.181,
2153
+ "step": 1700
2154
+ },
2155
+ {
2156
+ "epoch": 2.83,
2157
+ "learning_rate": 2.0862089632153988e-05,
2158
+ "loss": 0.1656,
2159
+ "step": 1705
2160
+ },
2161
+ {
2162
+ "epoch": 2.84,
2163
+ "learning_rate": 2.0729473722014436e-05,
2164
+ "loss": 0.1723,
2165
+ "step": 1710
2166
+ },
2167
+ {
2168
+ "epoch": 2.85,
2169
+ "learning_rate": 2.0596981478330748e-05,
2170
+ "loss": 0.1516,
2171
+ "step": 1715
2172
+ },
2173
+ {
2174
+ "epoch": 2.86,
2175
+ "learning_rate": 2.0464616737830567e-05,
2176
+ "loss": 0.1763,
2177
+ "step": 1720
2178
+ },
2179
+ {
2180
+ "epoch": 2.86,
2181
+ "learning_rate": 2.0332383333549288e-05,
2182
+ "loss": 0.1791,
2183
+ "step": 1725
2184
+ },
2185
+ {
2186
+ "epoch": 2.87,
2187
+ "learning_rate": 2.020028509471904e-05,
2188
+ "loss": 0.1577,
2189
+ "step": 1730
2190
+ },
2191
+ {
2192
+ "epoch": 2.88,
2193
+ "learning_rate": 2.006832584665783e-05,
2194
+ "loss": 0.1635,
2195
+ "step": 1735
2196
+ },
2197
+ {
2198
+ "epoch": 2.89,
2199
+ "learning_rate": 1.9936509410658723e-05,
2200
+ "loss": 0.1613,
2201
+ "step": 1740
2202
+ },
2203
+ {
2204
+ "epoch": 2.9,
2205
+ "learning_rate": 1.9804839603879238e-05,
2206
+ "loss": 0.1557,
2207
+ "step": 1745
2208
+ },
2209
+ {
2210
+ "epoch": 2.91,
2211
+ "learning_rate": 1.967332023923078e-05,
2212
+ "loss": 0.1732,
2213
+ "step": 1750
2214
+ },
2215
+ {
2216
+ "epoch": 2.91,
2217
+ "eval_loss": 0.09610436111688614,
2218
+ "eval_runtime": 6.569,
2219
+ "eval_samples_per_second": 15.223,
2220
+ "eval_steps_per_second": 15.223,
2221
+ "step": 1750
2222
+ },
2223
+ {
2224
+ "epoch": 2.91,
2225
+ "learning_rate": 1.9541955125268235e-05,
2226
+ "loss": 0.1803,
2227
+ "step": 1755
2228
+ },
2229
+ {
2230
+ "epoch": 2.92,
2231
+ "learning_rate": 1.941074806607968e-05,
2232
+ "loss": 0.1544,
2233
+ "step": 1760
2234
+ },
2235
+ {
2236
+ "epoch": 2.93,
2237
+ "learning_rate": 1.9279702861176202e-05,
2238
+ "loss": 0.1784,
2239
+ "step": 1765
2240
+ },
2241
+ {
2242
+ "epoch": 2.94,
2243
+ "learning_rate": 1.9148823305381898e-05,
2244
+ "loss": 0.1631,
2245
+ "step": 1770
2246
+ },
2247
+ {
2248
+ "epoch": 2.95,
2249
+ "learning_rate": 1.9018113188723998e-05,
2250
+ "loss": 0.1675,
2251
+ "step": 1775
2252
+ },
2253
+ {
2254
+ "epoch": 2.95,
2255
+ "learning_rate": 1.888757629632307e-05,
2256
+ "loss": 0.1702,
2257
+ "step": 1780
2258
+ },
2259
+ {
2260
+ "epoch": 2.96,
2261
+ "learning_rate": 1.875721640828344e-05,
2262
+ "loss": 0.1641,
2263
+ "step": 1785
2264
+ },
2265
+ {
2266
+ "epoch": 2.97,
2267
+ "learning_rate": 1.8627037299583703e-05,
2268
+ "loss": 0.1687,
2269
+ "step": 1790
2270
+ },
2271
+ {
2272
+ "epoch": 2.98,
2273
+ "learning_rate": 1.8497042739967445e-05,
2274
+ "loss": 0.1706,
2275
+ "step": 1795
2276
+ },
2277
+ {
2278
+ "epoch": 2.99,
2279
+ "learning_rate": 1.836723649383404e-05,
2280
+ "loss": 0.1575,
2281
+ "step": 1800
2282
+ },
2283
+ {
2284
+ "epoch": 3.0,
2285
+ "learning_rate": 1.8237622320129665e-05,
2286
+ "loss": 0.1608,
2287
+ "step": 1805
2288
+ },
2289
+ {
2290
+ "epoch": 3.0,
2291
+ "learning_rate": 1.8108203972238436e-05,
2292
+ "loss": 0.1506,
2293
+ "step": 1810
2294
+ },
2295
+ {
2296
+ "epoch": 3.01,
2297
+ "learning_rate": 1.797898519787372e-05,
2298
+ "loss": 0.134,
2299
+ "step": 1815
2300
+ },
2301
+ {
2302
+ "epoch": 3.02,
2303
+ "learning_rate": 1.784996973896959e-05,
2304
+ "loss": 0.148,
2305
+ "step": 1820
2306
+ },
2307
+ {
2308
+ "epoch": 3.03,
2309
+ "learning_rate": 1.7721161331572523e-05,
2310
+ "loss": 0.1457,
2311
+ "step": 1825
2312
+ },
2313
+ {
2314
+ "epoch": 3.04,
2315
+ "learning_rate": 1.759256370573313e-05,
2316
+ "loss": 0.1503,
2317
+ "step": 1830
2318
+ },
2319
+ {
2320
+ "epoch": 3.05,
2321
+ "learning_rate": 1.7464180585398205e-05,
2322
+ "loss": 0.142,
2323
+ "step": 1835
2324
+ },
2325
+ {
2326
+ "epoch": 3.05,
2327
+ "learning_rate": 1.733601568830287e-05,
2328
+ "loss": 0.1492,
2329
+ "step": 1840
2330
+ },
2331
+ {
2332
+ "epoch": 3.06,
2333
+ "learning_rate": 1.720807272586288e-05,
2334
+ "loss": 0.1662,
2335
+ "step": 1845
2336
+ },
2337
+ {
2338
+ "epoch": 3.07,
2339
+ "learning_rate": 1.708035540306721e-05,
2340
+ "loss": 0.1461,
2341
+ "step": 1850
2342
+ },
2343
+ {
2344
+ "epoch": 3.08,
2345
+ "learning_rate": 1.695286741837071e-05,
2346
+ "loss": 0.1496,
2347
+ "step": 1855
2348
+ },
2349
+ {
2350
+ "epoch": 3.09,
2351
+ "learning_rate": 1.6825612463587044e-05,
2352
+ "loss": 0.1448,
2353
+ "step": 1860
2354
+ },
2355
+ {
2356
+ "epoch": 3.1,
2357
+ "learning_rate": 1.669859422378176e-05,
2358
+ "loss": 0.1588,
2359
+ "step": 1865
2360
+ },
2361
+ {
2362
+ "epoch": 3.1,
2363
+ "learning_rate": 1.6571816377165567e-05,
2364
+ "loss": 0.153,
2365
+ "step": 1870
2366
+ },
2367
+ {
2368
+ "epoch": 3.11,
2369
+ "learning_rate": 1.6445282594987862e-05,
2370
+ "loss": 0.1511,
2371
+ "step": 1875
2372
+ },
2373
+ {
2374
+ "epoch": 3.11,
2375
+ "eval_loss": 0.0892980769276619,
2376
+ "eval_runtime": 6.5864,
2377
+ "eval_samples_per_second": 15.183,
2378
+ "eval_steps_per_second": 15.183,
2379
+ "step": 1875
2380
+ },
2381
+ {
2382
+ "epoch": 3.12,
2383
+ "learning_rate": 1.631899654143038e-05,
2384
+ "loss": 0.1455,
2385
+ "step": 1880
2386
+ },
2387
+ {
2388
+ "epoch": 3.13,
2389
+ "learning_rate": 1.6192961873501096e-05,
2390
+ "loss": 0.141,
2391
+ "step": 1885
2392
+ },
2393
+ {
2394
+ "epoch": 3.14,
2395
+ "learning_rate": 1.606718224092833e-05,
2396
+ "loss": 0.1423,
2397
+ "step": 1890
2398
+ },
2399
+ {
2400
+ "epoch": 3.15,
2401
+ "learning_rate": 1.5941661286055068e-05,
2402
+ "loss": 0.1583,
2403
+ "step": 1895
2404
+ },
2405
+ {
2406
+ "epoch": 3.15,
2407
+ "learning_rate": 1.581640264373344e-05,
2408
+ "loss": 0.138,
2409
+ "step": 1900
2410
+ },
2411
+ {
2412
+ "epoch": 3.16,
2413
+ "learning_rate": 1.5691409941219526e-05,
2414
+ "loss": 0.1543,
2415
+ "step": 1905
2416
+ },
2417
+ {
2418
+ "epoch": 3.17,
2419
+ "learning_rate": 1.556668679806828e-05,
2420
+ "loss": 0.1432,
2421
+ "step": 1910
2422
+ },
2423
+ {
2424
+ "epoch": 3.18,
2425
+ "learning_rate": 1.5442236826028706e-05,
2426
+ "loss": 0.146,
2427
+ "step": 1915
2428
+ },
2429
+ {
2430
+ "epoch": 3.19,
2431
+ "learning_rate": 1.531806362893932e-05,
2432
+ "loss": 0.1438,
2433
+ "step": 1920
2434
+ },
2435
+ {
2436
+ "epoch": 3.2,
2437
+ "learning_rate": 1.5194170802623691e-05,
2438
+ "loss": 0.1331,
2439
+ "step": 1925
2440
+ },
2441
+ {
2442
+ "epoch": 3.2,
2443
+ "learning_rate": 1.5070561934786437e-05,
2444
+ "loss": 0.1517,
2445
+ "step": 1930
2446
+ },
2447
+ {
2448
+ "epoch": 3.21,
2449
+ "learning_rate": 1.4947240604909232e-05,
2450
+ "loss": 0.1432,
2451
+ "step": 1935
2452
+ },
2453
+ {
2454
+ "epoch": 3.22,
2455
+ "learning_rate": 1.4824210384147207e-05,
2456
+ "loss": 0.1499,
2457
+ "step": 1940
2458
+ },
2459
+ {
2460
+ "epoch": 3.23,
2461
+ "learning_rate": 1.4701474835225515e-05,
2462
+ "loss": 0.1368,
2463
+ "step": 1945
2464
+ },
2465
+ {
2466
+ "epoch": 3.24,
2467
+ "learning_rate": 1.4579037512336138e-05,
2468
+ "loss": 0.1385,
2469
+ "step": 1950
2470
+ },
2471
+ {
2472
+ "epoch": 3.25,
2473
+ "learning_rate": 1.445690196103503e-05,
2474
+ "loss": 0.146,
2475
+ "step": 1955
2476
+ },
2477
+ {
2478
+ "epoch": 3.25,
2479
+ "learning_rate": 1.4335071718139378e-05,
2480
+ "loss": 0.1489,
2481
+ "step": 1960
2482
+ },
2483
+ {
2484
+ "epoch": 3.26,
2485
+ "learning_rate": 1.4213550311625216e-05,
2486
+ "loss": 0.1434,
2487
+ "step": 1965
2488
+ },
2489
+ {
2490
+ "epoch": 3.27,
2491
+ "learning_rate": 1.4092341260525248e-05,
2492
+ "loss": 0.1645,
2493
+ "step": 1970
2494
+ },
2495
+ {
2496
+ "epoch": 3.28,
2497
+ "learning_rate": 1.397144807482696e-05,
2498
+ "loss": 0.1562,
2499
+ "step": 1975
2500
+ },
2501
+ {
2502
+ "epoch": 3.29,
2503
+ "learning_rate": 1.385087425537096e-05,
2504
+ "loss": 0.1426,
2505
+ "step": 1980
2506
+ },
2507
+ {
2508
+ "epoch": 3.3,
2509
+ "learning_rate": 1.3730623293749606e-05,
2510
+ "loss": 0.1546,
2511
+ "step": 1985
2512
+ },
2513
+ {
2514
+ "epoch": 3.3,
2515
+ "learning_rate": 1.3610698672205895e-05,
2516
+ "loss": 0.1568,
2517
+ "step": 1990
2518
+ },
2519
+ {
2520
+ "epoch": 3.31,
2521
+ "learning_rate": 1.3491103863532626e-05,
2522
+ "loss": 0.1401,
2523
+ "step": 1995
2524
+ },
2525
+ {
2526
+ "epoch": 3.32,
2527
+ "learning_rate": 1.3371842330971862e-05,
2528
+ "loss": 0.1417,
2529
+ "step": 2000
2530
+ },
2531
+ {
2532
+ "epoch": 3.32,
2533
+ "eval_loss": 0.08844374865293503,
2534
+ "eval_runtime": 6.6133,
2535
+ "eval_samples_per_second": 15.121,
2536
+ "eval_steps_per_second": 15.121,
2537
+ "step": 2000
2538
+ },
2539
+ {
2540
+ "epoch": 3.33,
2541
+ "learning_rate": 1.3252917528114556e-05,
2542
+ "loss": 0.1392,
2543
+ "step": 2005
2544
+ },
2545
+ {
2546
+ "epoch": 3.34,
2547
+ "learning_rate": 1.3134332898800645e-05,
2548
+ "loss": 0.1457,
2549
+ "step": 2010
2550
+ },
2551
+ {
2552
+ "epoch": 3.35,
2553
+ "learning_rate": 1.3016091877019248e-05,
2554
+ "loss": 0.1524,
2555
+ "step": 2015
2556
+ },
2557
+ {
2558
+ "epoch": 3.35,
2559
+ "learning_rate": 1.2898197886809287e-05,
2560
+ "loss": 0.1538,
2561
+ "step": 2020
2562
+ },
2563
+ {
2564
+ "epoch": 3.36,
2565
+ "learning_rate": 1.2780654342160277e-05,
2566
+ "loss": 0.1491,
2567
+ "step": 2025
2568
+ },
2569
+ {
2570
+ "epoch": 3.37,
2571
+ "learning_rate": 1.2663464646913458e-05,
2572
+ "loss": 0.1402,
2573
+ "step": 2030
2574
+ },
2575
+ {
2576
+ "epoch": 3.38,
2577
+ "learning_rate": 1.2546632194663305e-05,
2578
+ "loss": 0.1525,
2579
+ "step": 2035
2580
+ },
2581
+ {
2582
+ "epoch": 3.39,
2583
+ "learning_rate": 1.2430160368659171e-05,
2584
+ "loss": 0.1396,
2585
+ "step": 2040
2586
+ },
2587
+ {
2588
+ "epoch": 3.39,
2589
+ "learning_rate": 1.2314052541707355e-05,
2590
+ "loss": 0.151,
2591
+ "step": 2045
2592
+ },
2593
+ {
2594
+ "epoch": 3.4,
2595
+ "learning_rate": 1.2198312076073426e-05,
2596
+ "loss": 0.1485,
2597
+ "step": 2050
2598
+ },
2599
+ {
2600
+ "epoch": 3.41,
2601
+ "learning_rate": 1.2082942323384852e-05,
2602
+ "loss": 0.1523,
2603
+ "step": 2055
2604
+ },
2605
+ {
2606
+ "epoch": 3.42,
2607
+ "learning_rate": 1.1967946624533948e-05,
2608
+ "loss": 0.141,
2609
+ "step": 2060
2610
+ },
2611
+ {
2612
+ "epoch": 3.43,
2613
+ "learning_rate": 1.1853328309581139e-05,
2614
+ "loss": 0.1451,
2615
+ "step": 2065
2616
+ },
2617
+ {
2618
+ "epoch": 3.44,
2619
+ "learning_rate": 1.1739090697658497e-05,
2620
+ "loss": 0.1636,
2621
+ "step": 2070
2622
+ },
2623
+ {
2624
+ "epoch": 3.44,
2625
+ "learning_rate": 1.162523709687367e-05,
2626
+ "loss": 0.1564,
2627
+ "step": 2075
2628
+ },
2629
+ {
2630
+ "epoch": 3.45,
2631
+ "learning_rate": 1.1511770804214044e-05,
2632
+ "loss": 0.143,
2633
+ "step": 2080
2634
+ },
2635
+ {
2636
+ "epoch": 3.46,
2637
+ "learning_rate": 1.1398695105451298e-05,
2638
+ "loss": 0.1379,
2639
+ "step": 2085
2640
+ },
2641
+ {
2642
+ "epoch": 3.47,
2643
+ "learning_rate": 1.1286013275046232e-05,
2644
+ "loss": 0.1386,
2645
+ "step": 2090
2646
+ },
2647
+ {
2648
+ "epoch": 3.48,
2649
+ "learning_rate": 1.1173728576053949e-05,
2650
+ "loss": 0.145,
2651
+ "step": 2095
2652
+ },
2653
+ {
2654
+ "epoch": 3.49,
2655
+ "learning_rate": 1.10618442600294e-05,
2656
+ "loss": 0.148,
2657
+ "step": 2100
2658
+ },
2659
+ {
2660
+ "epoch": 3.49,
2661
+ "learning_rate": 1.0950363566933172e-05,
2662
+ "loss": 0.1501,
2663
+ "step": 2105
2664
+ },
2665
+ {
2666
+ "epoch": 3.5,
2667
+ "learning_rate": 1.0839289725037668e-05,
2668
+ "loss": 0.1774,
2669
+ "step": 2110
2670
+ },
2671
+ {
2672
+ "epoch": 3.51,
2673
+ "learning_rate": 1.0728625950833665e-05,
2674
+ "loss": 0.1456,
2675
+ "step": 2115
2676
+ },
2677
+ {
2678
+ "epoch": 3.52,
2679
+ "learning_rate": 1.0618375448937157e-05,
2680
+ "loss": 0.1518,
2681
+ "step": 2120
2682
+ },
2683
+ {
2684
+ "epoch": 3.53,
2685
+ "learning_rate": 1.050854141199652e-05,
2686
+ "loss": 0.1581,
2687
+ "step": 2125
2688
+ },
2689
+ {
2690
+ "epoch": 3.53,
2691
+ "eval_loss": 0.08569944649934769,
2692
+ "eval_runtime": 6.6352,
2693
+ "eval_samples_per_second": 15.071,
2694
+ "eval_steps_per_second": 15.071,
2695
+ "step": 2125
2696
+ },
2697
+ {
2698
+ "epoch": 3.54,
2699
+ "learning_rate": 1.0399127020600103e-05,
2700
+ "loss": 0.1498,
2701
+ "step": 2130
2702
+ },
2703
+ {
2704
+ "epoch": 3.54,
2705
+ "learning_rate": 1.0290135443184068e-05,
2706
+ "loss": 0.145,
2707
+ "step": 2135
2708
+ },
2709
+ {
2710
+ "epoch": 3.55,
2711
+ "learning_rate": 1.018156983594073e-05,
2712
+ "loss": 0.1418,
2713
+ "step": 2140
2714
+ },
2715
+ {
2716
+ "epoch": 3.56,
2717
+ "learning_rate": 1.0073433342727069e-05,
2718
+ "loss": 0.1575,
2719
+ "step": 2145
2720
+ },
2721
+ {
2722
+ "epoch": 3.57,
2723
+ "learning_rate": 9.965729094973725e-06,
2724
+ "loss": 0.1474,
2725
+ "step": 2150
2726
+ },
2727
+ {
2728
+ "epoch": 3.58,
2729
+ "learning_rate": 9.85846021159433e-06,
2730
+ "loss": 0.1456,
2731
+ "step": 2155
2732
+ },
2733
+ {
2734
+ "epoch": 3.59,
2735
+ "learning_rate": 9.751629798895173e-06,
2736
+ "loss": 0.1369,
2737
+ "step": 2160
2738
+ },
2739
+ {
2740
+ "epoch": 3.59,
2741
+ "learning_rate": 9.645240950485248e-06,
2742
+ "loss": 0.1391,
2743
+ "step": 2165
2744
+ },
2745
+ {
2746
+ "epoch": 3.6,
2747
+ "learning_rate": 9.53929674718668e-06,
2748
+ "loss": 0.1429,
2749
+ "step": 2170
2750
+ },
2751
+ {
2752
+ "epoch": 3.61,
2753
+ "learning_rate": 9.433800256945499e-06,
2754
+ "loss": 0.1555,
2755
+ "step": 2175
2756
+ },
2757
+ {
2758
+ "epoch": 3.62,
2759
+ "learning_rate": 9.328754534742793e-06,
2760
+ "loss": 0.1435,
2761
+ "step": 2180
2762
+ },
2763
+ {
2764
+ "epoch": 3.63,
2765
+ "learning_rate": 9.224162622506283e-06,
2766
+ "loss": 0.1709,
2767
+ "step": 2185
2768
+ },
2769
+ {
2770
+ "epoch": 3.64,
2771
+ "learning_rate": 9.120027549022156e-06,
2772
+ "loss": 0.1467,
2773
+ "step": 2190
2774
+ },
2775
+ {
2776
+ "epoch": 3.64,
2777
+ "learning_rate": 9.016352329847433e-06,
2778
+ "loss": 0.1572,
2779
+ "step": 2195
2780
+ },
2781
+ {
2782
+ "epoch": 3.65,
2783
+ "learning_rate": 8.9131399672226e-06,
2784
+ "loss": 0.1394,
2785
+ "step": 2200
2786
+ },
2787
+ {
2788
+ "epoch": 3.66,
2789
+ "learning_rate": 8.810393449984705e-06,
2790
+ "loss": 0.1266,
2791
+ "step": 2205
2792
+ },
2793
+ {
2794
+ "epoch": 3.67,
2795
+ "learning_rate": 8.708115753480769e-06,
2796
+ "loss": 0.152,
2797
+ "step": 2210
2798
+ },
2799
+ {
2800
+ "epoch": 3.68,
2801
+ "learning_rate": 8.606309839481627e-06,
2802
+ "loss": 0.147,
2803
+ "step": 2215
2804
+ },
2805
+ {
2806
+ "epoch": 3.69,
2807
+ "learning_rate": 8.504978656096185e-06,
2808
+ "loss": 0.1555,
2809
+ "step": 2220
2810
+ },
2811
+ {
2812
+ "epoch": 3.69,
2813
+ "learning_rate": 8.404125137686061e-06,
2814
+ "loss": 0.1414,
2815
+ "step": 2225
2816
+ },
2817
+ {
2818
+ "epoch": 3.7,
2819
+ "learning_rate": 8.303752204780563e-06,
2820
+ "loss": 0.1362,
2821
+ "step": 2230
2822
+ },
2823
+ {
2824
+ "epoch": 3.71,
2825
+ "learning_rate": 8.203862763992153e-06,
2826
+ "loss": 0.142,
2827
+ "step": 2235
2828
+ },
2829
+ {
2830
+ "epoch": 3.72,
2831
+ "learning_rate": 8.104459707932237e-06,
2832
+ "loss": 0.1435,
2833
+ "step": 2240
2834
+ },
2835
+ {
2836
+ "epoch": 3.73,
2837
+ "learning_rate": 8.005545915127479e-06,
2838
+ "loss": 0.1547,
2839
+ "step": 2245
2840
+ },
2841
+ {
2842
+ "epoch": 3.74,
2843
+ "learning_rate": 7.907124249936356e-06,
2844
+ "loss": 0.1463,
2845
+ "step": 2250
2846
+ },
2847
+ {
2848
+ "epoch": 3.74,
2849
+ "eval_loss": 0.08357314765453339,
2850
+ "eval_runtime": 6.642,
2851
+ "eval_samples_per_second": 15.056,
2852
+ "eval_steps_per_second": 15.056,
2853
+ "step": 2250
2854
+ },
2855
+ {
2856
+ "epoch": 3.74,
2857
+ "learning_rate": 7.809197562466264e-06,
2858
+ "loss": 0.144,
2859
+ "step": 2255
2860
+ },
2861
+ {
2862
+ "epoch": 3.75,
2863
+ "learning_rate": 7.711768688490962e-06,
2864
+ "loss": 0.1512,
2865
+ "step": 2260
2866
+ },
2867
+ {
2868
+ "epoch": 3.76,
2869
+ "learning_rate": 7.614840449368469e-06,
2870
+ "loss": 0.156,
2871
+ "step": 2265
2872
+ },
2873
+ {
2874
+ "epoch": 3.77,
2875
+ "learning_rate": 7.518415651959349e-06,
2876
+ "loss": 0.1491,
2877
+ "step": 2270
2878
+ },
2879
+ {
2880
+ "epoch": 3.78,
2881
+ "learning_rate": 7.422497088545435e-06,
2882
+ "loss": 0.1508,
2883
+ "step": 2275
2884
+ },
2885
+ {
2886
+ "epoch": 3.79,
2887
+ "learning_rate": 7.32708753674897e-06,
2888
+ "loss": 0.1321,
2889
+ "step": 2280
2890
+ },
2891
+ {
2892
+ "epoch": 3.79,
2893
+ "learning_rate": 7.2321897594521875e-06,
2894
+ "loss": 0.1382,
2895
+ "step": 2285
2896
+ },
2897
+ {
2898
+ "epoch": 3.8,
2899
+ "learning_rate": 7.137806504717276e-06,
2900
+ "loss": 0.1567,
2901
+ "step": 2290
2902
+ },
2903
+ {
2904
+ "epoch": 3.81,
2905
+ "learning_rate": 7.0439405057068e-06,
2906
+ "loss": 0.1428,
2907
+ "step": 2295
2908
+ },
2909
+ {
2910
+ "epoch": 3.82,
2911
+ "learning_rate": 6.950594480604572e-06,
2912
+ "loss": 0.149,
2913
+ "step": 2300
2914
+ },
2915
+ {
2916
+ "epoch": 3.83,
2917
+ "learning_rate": 6.8577711325369584e-06,
2918
+ "loss": 0.1476,
2919
+ "step": 2305
2920
+ },
2921
+ {
2922
+ "epoch": 3.83,
2923
+ "learning_rate": 6.765473149494544e-06,
2924
+ "loss": 0.1463,
2925
+ "step": 2310
2926
+ },
2927
+ {
2928
+ "epoch": 3.84,
2929
+ "learning_rate": 6.673703204254347e-06,
2930
+ "loss": 0.1459,
2931
+ "step": 2315
2932
+ },
2933
+ {
2934
+ "epoch": 3.85,
2935
+ "learning_rate": 6.5824639543023674e-06,
2936
+ "loss": 0.1573,
2937
+ "step": 2320
2938
+ },
2939
+ {
2940
+ "epoch": 3.86,
2941
+ "learning_rate": 6.491758041756701e-06,
2942
+ "loss": 0.1415,
2943
+ "step": 2325
2944
+ },
2945
+ {
2946
+ "epoch": 3.87,
2947
+ "learning_rate": 6.401588093290964e-06,
2948
+ "loss": 0.1464,
2949
+ "step": 2330
2950
+ },
2951
+ {
2952
+ "epoch": 3.88,
2953
+ "learning_rate": 6.311956720058263e-06,
2954
+ "loss": 0.1644,
2955
+ "step": 2335
2956
+ },
2957
+ {
2958
+ "epoch": 3.88,
2959
+ "learning_rate": 6.222866517615569e-06,
2960
+ "loss": 0.145,
2961
+ "step": 2340
2962
+ },
2963
+ {
2964
+ "epoch": 3.89,
2965
+ "learning_rate": 6.1343200658485645e-06,
2966
+ "loss": 0.1371,
2967
+ "step": 2345
2968
+ },
2969
+ {
2970
+ "epoch": 3.9,
2971
+ "learning_rate": 6.046319928896926e-06,
2972
+ "loss": 0.1406,
2973
+ "step": 2350
2974
+ },
2975
+ {
2976
+ "epoch": 3.91,
2977
+ "learning_rate": 5.958868655080077e-06,
2978
+ "loss": 0.1483,
2979
+ "step": 2355
2980
+ },
2981
+ {
2982
+ "epoch": 3.92,
2983
+ "learning_rate": 5.871968776823383e-06,
2984
+ "loss": 0.1397,
2985
+ "step": 2360
2986
+ },
2987
+ {
2988
+ "epoch": 3.93,
2989
+ "learning_rate": 5.785622810584834e-06,
2990
+ "loss": 0.1467,
2991
+ "step": 2365
2992
+ },
2993
+ {
2994
+ "epoch": 3.93,
2995
+ "learning_rate": 5.699833256782161e-06,
2996
+ "loss": 0.1545,
2997
+ "step": 2370
2998
+ },
2999
+ {
3000
+ "epoch": 3.94,
3001
+ "learning_rate": 5.614602599720431e-06,
3002
+ "loss": 0.1487,
3003
+ "step": 2375
3004
+ },
3005
+ {
3006
+ "epoch": 3.94,
3007
+ "eval_loss": 0.08115073293447495,
3008
+ "eval_runtime": 6.6449,
3009
+ "eval_samples_per_second": 15.049,
3010
+ "eval_steps_per_second": 15.049,
3011
+ "step": 2375
3012
+ },
3013
+ {
3014
+ "epoch": 3.95,
3015
+ "learning_rate": 5.529933307520102e-06,
3016
+ "loss": 0.156,
3017
+ "step": 2380
3018
+ },
3019
+ {
3020
+ "epoch": 3.96,
3021
+ "learning_rate": 5.445827832045558e-06,
3022
+ "loss": 0.1516,
3023
+ "step": 2385
3024
+ },
3025
+ {
3026
+ "epoch": 3.97,
3027
+ "learning_rate": 5.362288608834115e-06,
3028
+ "loss": 0.1514,
3029
+ "step": 2390
3030
+ },
3031
+ {
3032
+ "epoch": 3.98,
3033
+ "learning_rate": 5.2793180570254755e-06,
3034
+ "loss": 0.1449,
3035
+ "step": 2395
3036
+ },
3037
+ {
3038
+ "epoch": 3.98,
3039
+ "learning_rate": 5.196918579291668e-06,
3040
+ "loss": 0.1507,
3041
+ "step": 2400
3042
+ },
3043
+ {
3044
+ "epoch": 3.99,
3045
+ "learning_rate": 5.11509256176749e-06,
3046
+ "loss": 0.1579,
3047
+ "step": 2405
3048
+ },
3049
+ {
3050
+ "epoch": 4.0,
3051
+ "learning_rate": 5.0338423739814285e-06,
3052
+ "loss": 0.1376,
3053
+ "step": 2410
3054
+ },
3055
+ {
3056
+ "epoch": 4.01,
3057
+ "learning_rate": 4.9531703687869855e-06,
3058
+ "loss": 0.1462,
3059
+ "step": 2415
3060
+ },
3061
+ {
3062
+ "epoch": 4.02,
3063
+ "learning_rate": 4.873078882294599e-06,
3064
+ "loss": 0.1319,
3065
+ "step": 2420
3066
+ },
3067
+ {
3068
+ "epoch": 4.03,
3069
+ "learning_rate": 4.79357023380394e-06,
3070
+ "loss": 0.1369,
3071
+ "step": 2425
3072
+ },
3073
+ {
3074
+ "epoch": 4.03,
3075
+ "learning_rate": 4.714646725736821e-06,
3076
+ "loss": 0.1475,
3077
+ "step": 2430
3078
+ },
3079
+ {
3080
+ "epoch": 4.04,
3081
+ "learning_rate": 4.636310643570454e-06,
3082
+ "loss": 0.1369,
3083
+ "step": 2435
3084
+ },
3085
+ {
3086
+ "epoch": 4.05,
3087
+ "learning_rate": 4.5585642557713085e-06,
3088
+ "loss": 0.1303,
3089
+ "step": 2440
3090
+ },
3091
+ {
3092
+ "epoch": 4.06,
3093
+ "learning_rate": 4.481409813729398e-06,
3094
+ "loss": 0.1395,
3095
+ "step": 2445
3096
+ },
3097
+ {
3098
+ "epoch": 4.07,
3099
+ "learning_rate": 4.404849551693102e-06,
3100
+ "loss": 0.1342,
3101
+ "step": 2450
3102
+ },
3103
+ {
3104
+ "epoch": 4.08,
3105
+ "learning_rate": 4.328885686704456e-06,
3106
+ "loss": 0.1372,
3107
+ "step": 2455
3108
+ },
3109
+ {
3110
+ "epoch": 4.08,
3111
+ "learning_rate": 4.2535204185349455e-06,
3112
+ "loss": 0.1436,
3113
+ "step": 2460
3114
+ },
3115
+ {
3116
+ "epoch": 4.09,
3117
+ "learning_rate": 4.178755929621822e-06,
3118
+ "loss": 0.129,
3119
+ "step": 2465
3120
+ },
3121
+ {
3122
+ "epoch": 4.1,
3123
+ "learning_rate": 4.104594385004884e-06,
3124
+ "loss": 0.1378,
3125
+ "step": 2470
3126
+ },
3127
+ {
3128
+ "epoch": 4.11,
3129
+ "learning_rate": 4.031037932263798e-06,
3130
+ "loss": 0.1472,
3131
+ "step": 2475
3132
+ },
3133
+ {
3134
+ "epoch": 4.12,
3135
+ "learning_rate": 3.958088701455895e-06,
3136
+ "loss": 0.1155,
3137
+ "step": 2480
3138
+ },
3139
+ {
3140
+ "epoch": 4.13,
3141
+ "learning_rate": 3.88574880505449e-06,
3142
+ "loss": 0.1397,
3143
+ "step": 2485
3144
+ },
3145
+ {
3146
+ "epoch": 4.13,
3147
+ "learning_rate": 3.814020337887736e-06,
3148
+ "loss": 0.1313,
3149
+ "step": 2490
3150
+ },
3151
+ {
3152
+ "epoch": 4.14,
3153
+ "learning_rate": 3.7429053770779127e-06,
3154
+ "loss": 0.1342,
3155
+ "step": 2495
3156
+ },
3157
+ {
3158
+ "epoch": 4.15,
3159
+ "learning_rate": 3.672405981981322e-06,
3160
+ "loss": 0.1453,
3161
+ "step": 2500
3162
+ },
3163
+ {
3164
+ "epoch": 4.15,
3165
+ "eval_loss": 0.07911655306816101,
3166
+ "eval_runtime": 6.6443,
3167
+ "eval_samples_per_second": 15.051,
3168
+ "eval_steps_per_second": 15.051,
3169
+ "step": 2500
3170
+ },
3171
+ {
3172
+ "epoch": 4.16,
3173
+ "learning_rate": 3.602524194128601e-06,
3174
+ "loss": 0.1416,
3175
+ "step": 2505
3176
+ },
3177
+ {
3178
+ "epoch": 4.17,
3179
+ "learning_rate": 3.5332620371656826e-06,
3180
+ "loss": 0.1426,
3181
+ "step": 2510
3182
+ },
3183
+ {
3184
+ "epoch": 4.18,
3185
+ "learning_rate": 3.464621516795116e-06,
3186
+ "loss": 0.148,
3187
+ "step": 2515
3188
+ },
3189
+ {
3190
+ "epoch": 4.18,
3191
+ "learning_rate": 3.396604620718025e-06,
3192
+ "loss": 0.1457,
3193
+ "step": 2520
3194
+ },
3195
+ {
3196
+ "epoch": 4.19,
3197
+ "learning_rate": 3.3292133185765407e-06,
3198
+ "loss": 0.1431,
3199
+ "step": 2525
3200
+ },
3201
+ {
3202
+ "epoch": 4.2,
3203
+ "learning_rate": 3.262449561896766e-06,
3204
+ "loss": 0.1398,
3205
+ "step": 2530
3206
+ },
3207
+ {
3208
+ "epoch": 4.21,
3209
+ "learning_rate": 3.19631528403225e-06,
3210
+ "loss": 0.1401,
3211
+ "step": 2535
3212
+ },
3213
+ {
3214
+ "epoch": 4.22,
3215
+ "learning_rate": 3.1308124001080296e-06,
3216
+ "loss": 0.1478,
3217
+ "step": 2540
3218
+ },
3219
+ {
3220
+ "epoch": 4.22,
3221
+ "learning_rate": 3.0659428069651388e-06,
3222
+ "loss": 0.1401,
3223
+ "step": 2545
3224
+ },
3225
+ {
3226
+ "epoch": 4.23,
3227
+ "learning_rate": 3.0017083831057004e-06,
3228
+ "loss": 0.1376,
3229
+ "step": 2550
3230
+ },
3231
+ {
3232
+ "epoch": 4.24,
3233
+ "learning_rate": 2.9381109886385205e-06,
3234
+ "loss": 0.1448,
3235
+ "step": 2555
3236
+ },
3237
+ {
3238
+ "epoch": 4.25,
3239
+ "learning_rate": 2.8751524652252335e-06,
3240
+ "loss": 0.1431,
3241
+ "step": 2560
3242
+ },
3243
+ {
3244
+ "epoch": 4.26,
3245
+ "learning_rate": 2.812834636026948e-06,
3246
+ "loss": 0.1484,
3247
+ "step": 2565
3248
+ },
3249
+ {
3250
+ "epoch": 4.27,
3251
+ "learning_rate": 2.751159305651477e-06,
3252
+ "loss": 0.1597,
3253
+ "step": 2570
3254
+ },
3255
+ {
3256
+ "epoch": 4.27,
3257
+ "learning_rate": 2.6901282601010687e-06,
3258
+ "loss": 0.1329,
3259
+ "step": 2575
3260
+ },
3261
+ {
3262
+ "epoch": 4.28,
3263
+ "learning_rate": 2.6297432667206912e-06,
3264
+ "loss": 0.1423,
3265
+ "step": 2580
3266
+ },
3267
+ {
3268
+ "epoch": 4.29,
3269
+ "learning_rate": 2.5700060741468274e-06,
3270
+ "loss": 0.1391,
3271
+ "step": 2585
3272
+ },
3273
+ {
3274
+ "epoch": 4.3,
3275
+ "learning_rate": 2.5109184122568795e-06,
3276
+ "loss": 0.1423,
3277
+ "step": 2590
3278
+ },
3279
+ {
3280
+ "epoch": 4.31,
3281
+ "learning_rate": 2.4524819921190654e-06,
3282
+ "loss": 0.1452,
3283
+ "step": 2595
3284
+ },
3285
+ {
3286
+ "epoch": 4.32,
3287
+ "learning_rate": 2.394698505942833e-06,
3288
+ "loss": 0.1364,
3289
+ "step": 2600
3290
+ },
3291
+ {
3292
+ "epoch": 4.32,
3293
+ "learning_rate": 2.337569627029909e-06,
3294
+ "loss": 0.1422,
3295
+ "step": 2605
3296
+ },
3297
+ {
3298
+ "epoch": 4.33,
3299
+ "learning_rate": 2.2810970097257913e-06,
3300
+ "loss": 0.1406,
3301
+ "step": 2610
3302
+ },
3303
+ {
3304
+ "epoch": 4.34,
3305
+ "learning_rate": 2.2252822893718912e-06,
3306
+ "loss": 0.1397,
3307
+ "step": 2615
3308
+ },
3309
+ {
3310
+ "epoch": 4.35,
3311
+ "learning_rate": 2.1701270822581472e-06,
3312
+ "loss": 0.1281,
3313
+ "step": 2620
3314
+ },
3315
+ {
3316
+ "epoch": 4.36,
3317
+ "learning_rate": 2.115632985576224e-06,
3318
+ "loss": 0.1457,
3319
+ "step": 2625
3320
+ },
3321
+ {
3322
+ "epoch": 4.36,
3323
+ "eval_loss": 0.07916607707738876,
3324
+ "eval_runtime": 6.6471,
3325
+ "eval_samples_per_second": 15.044,
3326
+ "eval_steps_per_second": 15.044,
3327
+ "step": 2625
3328
+ },
3329
+ {
3330
+ "epoch": 4.37,
3331
+ "learning_rate": 2.061801577373268e-06,
3332
+ "loss": 0.1481,
3333
+ "step": 2630
3334
+ },
3335
+ {
3336
+ "epoch": 4.37,
3337
+ "learning_rate": 2.0086344165062056e-06,
3338
+ "loss": 0.1433,
3339
+ "step": 2635
3340
+ },
3341
+ {
3342
+ "epoch": 4.38,
3343
+ "learning_rate": 1.956133042596603e-06,
3344
+ "loss": 0.1302,
3345
+ "step": 2640
3346
+ },
3347
+ {
3348
+ "epoch": 4.39,
3349
+ "learning_rate": 1.904298975986077e-06,
3350
+ "loss": 0.1364,
3351
+ "step": 2645
3352
+ },
3353
+ {
3354
+ "epoch": 4.4,
3355
+ "learning_rate": 1.853133717692279e-06,
3356
+ "loss": 0.146,
3357
+ "step": 2650
3358
+ },
3359
+ {
3360
+ "epoch": 4.41,
3361
+ "learning_rate": 1.8026387493654206e-06,
3362
+ "loss": 0.1277,
3363
+ "step": 2655
3364
+ },
3365
+ {
3366
+ "epoch": 4.42,
3367
+ "learning_rate": 1.752815533245364e-06,
3368
+ "loss": 0.1367,
3369
+ "step": 2660
3370
+ },
3371
+ {
3372
+ "epoch": 4.42,
3373
+ "learning_rate": 1.7036655121192874e-06,
3374
+ "loss": 0.1422,
3375
+ "step": 2665
3376
+ },
3377
+ {
3378
+ "epoch": 4.43,
3379
+ "learning_rate": 1.6551901092799027e-06,
3380
+ "loss": 0.1442,
3381
+ "step": 2670
3382
+ },
3383
+ {
3384
+ "epoch": 4.44,
3385
+ "learning_rate": 1.6073907284842315e-06,
3386
+ "loss": 0.135,
3387
+ "step": 2675
3388
+ },
3389
+ {
3390
+ "epoch": 4.45,
3391
+ "learning_rate": 1.5602687539129746e-06,
3392
+ "loss": 0.1372,
3393
+ "step": 2680
3394
+ },
3395
+ {
3396
+ "epoch": 4.46,
3397
+ "learning_rate": 1.5138255501304015e-06,
3398
+ "loss": 0.1296,
3399
+ "step": 2685
3400
+ },
3401
+ {
3402
+ "epoch": 4.47,
3403
+ "learning_rate": 1.4680624620448419e-06,
3404
+ "loss": 0.1421,
3405
+ "step": 2690
3406
+ },
3407
+ {
3408
+ "epoch": 4.47,
3409
+ "learning_rate": 1.4229808148697733e-06,
3410
+ "loss": 0.1319,
3411
+ "step": 2695
3412
+ },
3413
+ {
3414
+ "epoch": 4.48,
3415
+ "learning_rate": 1.3785819140853973e-06,
3416
+ "loss": 0.1338,
3417
+ "step": 2700
3418
+ },
3419
+ {
3420
+ "epoch": 4.49,
3421
+ "learning_rate": 1.3348670454008589e-06,
3422
+ "loss": 0.1387,
3423
+ "step": 2705
3424
+ },
3425
+ {
3426
+ "epoch": 4.5,
3427
+ "learning_rate": 1.2918374747170225e-06,
3428
+ "loss": 0.1414,
3429
+ "step": 2710
3430
+ },
3431
+ {
3432
+ "epoch": 4.51,
3433
+ "learning_rate": 1.249494448089794e-06,
3434
+ "loss": 0.1473,
3435
+ "step": 2715
3436
+ },
3437
+ {
3438
+ "epoch": 4.52,
3439
+ "learning_rate": 1.207839191694049e-06,
3440
+ "loss": 0.1453,
3441
+ "step": 2720
3442
+ },
3443
+ {
3444
+ "epoch": 4.52,
3445
+ "learning_rate": 1.1668729117881249e-06,
3446
+ "loss": 0.13,
3447
+ "step": 2725
3448
+ },
3449
+ {
3450
+ "epoch": 4.53,
3451
+ "learning_rate": 1.1265967946788914e-06,
3452
+ "loss": 0.1335,
3453
+ "step": 2730
3454
+ },
3455
+ {
3456
+ "epoch": 4.54,
3457
+ "learning_rate": 1.0870120066873878e-06,
3458
+ "loss": 0.1368,
3459
+ "step": 2735
3460
+ },
3461
+ {
3462
+ "epoch": 4.55,
3463
+ "learning_rate": 1.0481196941150572e-06,
3464
+ "loss": 0.139,
3465
+ "step": 2740
3466
+ },
3467
+ {
3468
+ "epoch": 4.56,
3469
+ "learning_rate": 1.0099209832105521e-06,
3470
+ "loss": 0.1417,
3471
+ "step": 2745
3472
+ },
3473
+ {
3474
+ "epoch": 4.57,
3475
+ "learning_rate": 9.724169801371168e-07,
3476
+ "loss": 0.1429,
3477
+ "step": 2750
3478
+ },
3479
+ {
3480
+ "epoch": 4.57,
3481
+ "eval_loss": 0.07882240414619446,
3482
+ "eval_runtime": 6.6515,
3483
+ "eval_samples_per_second": 15.034,
3484
+ "eval_steps_per_second": 15.034,
3485
+ "step": 2750
3486
+ },
3487
+ {
3488
+ "epoch": 4.57,
3489
+ "learning_rate": 9.356087709405465e-07,
3490
+ "loss": 0.1244,
3491
+ "step": 2755
3492
+ },
3493
+ {
3494
+ "epoch": 4.58,
3495
+ "learning_rate": 8.994974215177642e-07,
3496
+ "loss": 0.1291,
3497
+ "step": 2760
3498
+ },
3499
+ {
3500
+ "epoch": 4.59,
3501
+ "learning_rate": 8.640839775859221e-07,
3502
+ "loss": 0.1393,
3503
+ "step": 2765
3504
+ },
3505
+ {
3506
+ "epoch": 4.6,
3507
+ "learning_rate": 8.293694646521411e-07,
3508
+ "loss": 0.1441,
3509
+ "step": 2770
3510
+ },
3511
+ {
3512
+ "epoch": 4.61,
3513
+ "learning_rate": 7.953548879838013e-07,
3514
+ "loss": 0.1438,
3515
+ "step": 2775
3516
+ },
3517
+ {
3518
+ "epoch": 4.62,
3519
+ "learning_rate": 7.620412325794468e-07,
3520
+ "loss": 0.1533,
3521
+ "step": 2780
3522
+ },
3523
+ {
3524
+ "epoch": 4.62,
3525
+ "learning_rate": 7.294294631402493e-07,
3526
+ "loss": 0.1251,
3527
+ "step": 2785
3528
+ },
3529
+ {
3530
+ "epoch": 4.63,
3531
+ "learning_rate": 6.975205240420696e-07,
3532
+ "loss": 0.1355,
3533
+ "step": 2790
3534
+ },
3535
+ {
3536
+ "epoch": 4.64,
3537
+ "learning_rate": 6.663153393081184e-07,
3538
+ "loss": 0.1264,
3539
+ "step": 2795
3540
+ },
3541
+ {
3542
+ "epoch": 4.65,
3543
+ "learning_rate": 6.358148125822e-07,
3544
+ "loss": 0.1417,
3545
+ "step": 2800
3546
+ },
3547
+ {
3548
+ "epoch": 4.66,
3549
+ "learning_rate": 6.060198271025358e-07,
3550
+ "loss": 0.1311,
3551
+ "step": 2805
3552
+ },
3553
+ {
3554
+ "epoch": 4.66,
3555
+ "learning_rate": 5.769312456761989e-07,
3556
+ "loss": 0.1433,
3557
+ "step": 2810
3558
+ },
3559
+ {
3560
+ "epoch": 4.67,
3561
+ "learning_rate": 5.485499106541086e-07,
3562
+ "loss": 0.1417,
3563
+ "step": 2815
3564
+ },
3565
+ {
3566
+ "epoch": 4.68,
3567
+ "learning_rate": 5.208766439066676e-07,
3568
+ "loss": 0.1364,
3569
+ "step": 2820
3570
+ },
3571
+ {
3572
+ "epoch": 4.69,
3573
+ "learning_rate": 4.939122467999297e-07,
3574
+ "loss": 0.136,
3575
+ "step": 2825
3576
+ },
3577
+ {
3578
+ "epoch": 4.7,
3579
+ "learning_rate": 4.6765750017242206e-07,
3580
+ "loss": 0.1257,
3581
+ "step": 2830
3582
+ },
3583
+ {
3584
+ "epoch": 4.71,
3585
+ "learning_rate": 4.421131643125104e-07,
3586
+ "loss": 0.1329,
3587
+ "step": 2835
3588
+ },
3589
+ {
3590
+ "epoch": 4.71,
3591
+ "learning_rate": 4.172799789363996e-07,
3592
+ "loss": 0.1283,
3593
+ "step": 2840
3594
+ },
3595
+ {
3596
+ "epoch": 4.72,
3597
+ "learning_rate": 3.931586631667039e-07,
3598
+ "loss": 0.144,
3599
+ "step": 2845
3600
+ },
3601
+ {
3602
+ "epoch": 4.73,
3603
+ "learning_rate": 3.697499155116302e-07,
3604
+ "loss": 0.1349,
3605
+ "step": 2850
3606
+ },
3607
+ {
3608
+ "epoch": 4.74,
3609
+ "learning_rate": 3.4705441384473293e-07,
3610
+ "loss": 0.1394,
3611
+ "step": 2855
3612
+ },
3613
+ {
3614
+ "epoch": 4.75,
3615
+ "learning_rate": 3.25072815385305e-07,
3616
+ "loss": 0.1388,
3617
+ "step": 2860
3618
+ },
3619
+ {
3620
+ "epoch": 4.76,
3621
+ "learning_rate": 3.0380575667934285e-07,
3622
+ "loss": 0.1254,
3623
+ "step": 2865
3624
+ },
3625
+ {
3626
+ "epoch": 4.76,
3627
+ "learning_rate": 2.8325385358109467e-07,
3628
+ "loss": 0.1394,
3629
+ "step": 2870
3630
+ },
3631
+ {
3632
+ "epoch": 4.77,
3633
+ "learning_rate": 2.6341770123524676e-07,
3634
+ "loss": 0.1376,
3635
+ "step": 2875
3636
+ },
3637
+ {
3638
+ "epoch": 4.77,
3639
+ "eval_loss": 0.07855534553527832,
3640
+ "eval_runtime": 6.664,
3641
+ "eval_samples_per_second": 15.006,
3642
+ "eval_steps_per_second": 15.006,
3643
+ "step": 2875
3644
+ },
3645
+ {
3646
+ "epoch": 4.78,
3647
+ "learning_rate": 2.442978740596791e-07,
3648
+ "loss": 0.1286,
3649
+ "step": 2880
3650
+ },
3651
+ {
3652
+ "epoch": 4.79,
3653
+ "learning_rate": 2.258949257288423e-07,
3654
+ "loss": 0.1273,
3655
+ "step": 2885
3656
+ },
3657
+ {
3658
+ "epoch": 4.8,
3659
+ "learning_rate": 2.0820938915770416e-07,
3660
+ "loss": 0.1382,
3661
+ "step": 2890
3662
+ },
3663
+ {
3664
+ "epoch": 4.81,
3665
+ "learning_rate": 1.9124177648633667e-07,
3666
+ "loss": 0.1438,
3667
+ "step": 2895
3668
+ },
3669
+ {
3670
+ "epoch": 4.81,
3671
+ "learning_rate": 1.7499257906507526e-07,
3672
+ "loss": 0.1409,
3673
+ "step": 2900
3674
+ },
3675
+ {
3676
+ "epoch": 4.82,
3677
+ "learning_rate": 1.5946226744029402e-07,
3678
+ "loss": 0.1347,
3679
+ "step": 2905
3680
+ },
3681
+ {
3682
+ "epoch": 4.83,
3683
+ "learning_rate": 1.4465129134078048e-07,
3684
+ "loss": 0.1368,
3685
+ "step": 2910
3686
+ },
3687
+ {
3688
+ "epoch": 4.84,
3689
+ "learning_rate": 1.3056007966470717e-07,
3690
+ "loss": 0.1292,
3691
+ "step": 2915
3692
+ },
3693
+ {
3694
+ "epoch": 4.85,
3695
+ "learning_rate": 1.1718904046721658e-07,
3696
+ "loss": 0.1307,
3697
+ "step": 2920
3698
+ },
3699
+ {
3700
+ "epoch": 4.86,
3701
+ "learning_rate": 1.045385609486027e-07,
3702
+ "loss": 0.1525,
3703
+ "step": 2925
3704
+ },
3705
+ {
3706
+ "epoch": 4.86,
3707
+ "learning_rate": 9.260900744309797e-08,
3708
+ "loss": 0.1287,
3709
+ "step": 2930
3710
+ },
3711
+ {
3712
+ "epoch": 4.87,
3713
+ "learning_rate": 8.140072540826771e-08,
3714
+ "loss": 0.1312,
3715
+ "step": 2935
3716
+ },
3717
+ {
3718
+ "epoch": 4.88,
3719
+ "learning_rate": 7.091403941499598e-08,
3720
+ "loss": 0.1335,
3721
+ "step": 2940
3722
+ },
3723
+ {
3724
+ "epoch": 4.89,
3725
+ "learning_rate": 6.114925313810693e-08,
3726
+ "loss": 0.1395,
3727
+ "step": 2945
3728
+ },
3729
+ {
3730
+ "epoch": 4.9,
3731
+ "learning_rate": 5.210664934754972e-08,
3732
+ "loss": 0.1405,
3733
+ "step": 2950
3734
+ },
3735
+ {
3736
+ "epoch": 4.91,
3737
+ "learning_rate": 4.378648990022438e-08,
3738
+ "loss": 0.1343,
3739
+ "step": 2955
3740
+ },
3741
+ {
3742
+ "epoch": 4.91,
3743
+ "learning_rate": 3.618901573238798e-08,
3744
+ "loss": 0.1409,
3745
+ "step": 2960
3746
+ },
3747
+ {
3748
+ "epoch": 4.92,
3749
+ "learning_rate": 2.931444685269069e-08,
3750
+ "loss": 0.1327,
3751
+ "step": 2965
3752
+ },
3753
+ {
3754
+ "epoch": 4.93,
3755
+ "learning_rate": 2.3162982335789286e-08,
3756
+ "loss": 0.1367,
3757
+ "step": 2970
3758
+ },
3759
+ {
3760
+ "epoch": 4.94,
3761
+ "learning_rate": 1.7734800316596135e-08,
3762
+ "loss": 0.1259,
3763
+ "step": 2975
3764
+ },
3765
+ {
3766
+ "epoch": 4.95,
3767
+ "learning_rate": 1.3030057985108369e-08,
3768
+ "loss": 0.1325,
3769
+ "step": 2980
3770
+ },
3771
+ {
3772
+ "epoch": 4.96,
3773
+ "learning_rate": 9.048891581875385e-09,
3774
+ "loss": 0.1275,
3775
+ "step": 2985
3776
+ },
3777
+ {
3778
+ "epoch": 4.96,
3779
+ "learning_rate": 5.791416394024252e-09,
3780
+ "loss": 0.1253,
3781
+ "step": 2990
3782
+ },
3783
+ {
3784
+ "epoch": 4.97,
3785
+ "learning_rate": 3.2577267519484643e-09,
3786
+ "loss": 0.1408,
3787
+ "step": 2995
3788
+ },
3789
+ {
3790
+ "epoch": 4.98,
3791
+ "learning_rate": 1.4478960265629227e-09,
3792
+ "loss": 0.1278,
3793
+ "step": 3000
3794
+ },
3795
+ {
3796
+ "epoch": 4.98,
3797
+ "eval_loss": 0.07845821231603622,
3798
+ "eval_runtime": 6.6657,
3799
+ "eval_samples_per_second": 15.002,
3800
+ "eval_steps_per_second": 15.002,
3801
+ "step": 3000
3802
+ }
3803
+ ],
3804
+ "logging_steps": 5,
3805
+ "max_steps": 3010,
3806
+ "num_input_tokens_seen": 0,
3807
+ "num_train_epochs": 5,
3808
+ "save_steps": 125,
3809
+ "total_flos": 1.0266248056637276e+18,
3810
+ "train_batch_size": 1,
3811
+ "trial_name": null,
3812
+ "trial_params": null
3813
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4af4114624d731e2e2ff19b78348a21d5dc85a8f9fd9ccab538384c45dd93f6
3
+ size 4984